Skip to main content

Full text of "Measuring performance in managed care plans : final report"

See other formats


n he. 1IW 

HEALTH ECONOMICS RESEARCH, INC. 



41 1 Waverley Oaks Road, Suite 330 
Waltham, MA 02452 
(781)788-8100 
(781) 788-8101 -FAX 



MEASURING PERFORMANCE IN 
MANAGED CARE PLANS 



Final Report 



Submitted by : 



Debra A. Dayhoff, Ph.D. 

Margo L. Rosenbach, Ph.D. 

Edith G. Walsh, Ph.D. 



With : 

Health Economics Research, Inc. 

And : 

Mary B. Barton, M.D., M.P.P 

Robert H. Fletcher, M.D., M.Sc. 

Stephen B. Soumerai, Sc.D. 

With : 

Harvard Pilgrim Health Care 

September 25, 1998 



. 









Table of Contents 

Page 

Executive Summary ES-1 

Chapter 1 Introduction 1_1 

1.1 Objectives of the Project 1-1 

1.2 Monitoring Performance in Managed Care and Fee-for-Service 1-4 

1 .3 Defining Access in a Managed Care Environment 1-5 

1 .4 Conceptual Framework 1-7 

1 .5 Selection of Indicators 1-9 

1 .6 Indicators Included in this Study 1-11 

1 .7 Analysis of Disenrollees 1-13 

1.8 Organization of the Report 1-14 

Chapter 2 Methods 2-1 

2.1 Data Sources 2-1 

2.2 Sample Selection 2-4 

2.3 Constructing Annual Indicators 2-8 

2.4 Adjusted vs. Unadjusted Rates 2-9 

2.5 Comparison with External Benchmarks 2-10 

Chapter 3 Results 3-1 

3.1 Preventive Care 3-1 

3.2 Chronic Disease Care 3-12 

3.3 Diagnosis Specific Care 3-36 

3.4 Specialty Referral Care 3-44 

3.5 Primary Care 3-51 

3.6 Analysis of Disenrollees 3-59 

Chapter 4 Discussion 4-1 

4.1 Comparison of Fee-For-Service and Managed Care Results 4-1 

4.2 Implications for Developing a Monitoring System 4-6 

4.3 Conclusion 4-16 

References 



Health Economics Research, Inc. Table of Contents: i 

crimson\final\TOC.wpt\dpb 



Table of Contents 
(continued) 

Page 

Appendix A Profile of Harvard Pilgrim Health Care A-l 

1 .0 Overview A.1 

2.0 Membership and Enrollment A-5 

3.0 Benefits A-9 

4.0 Capacity and Service Delivery A-l 1 

5.0 Medical Management Systems A-14 

6.0 Mechanisms to Monitor and Promote Access A- 18 

7.0 Measurement of Member Satisfaction A-23 

Appendix B Sample Sizes for Access Indicators B-l 

Appendix C HPHC Data Descriptions 



Health Economics Research, Inc. Table of Contents: ii 

crimson\final\TOC.wpt\dpb 



Table of Tables, Figures, and Exhibits 



Page 



Executive Summary 



Table ES-1 Age- Adjusted Summary of Performance Indicators ES-6 

Chapter 1 

Exhibitl-1 Monitoring Delivery of Timely and Appropriate Care: 1-8 

Conceptual Framework 
Table 1-1 Summary of Medicare Performance Indicators 1-12 

Chapter 2 

Exhibit 2-1 Zip Codes in which HPHC Medicare Beneficiaries Reside 2-7 

Chapter 3 

Table 3-1 Age-Adjusted Summary of Performance Indicators 3-2 

Figure 3-1 Proportion of New Enrollees With a Visit at 30 Day 3-53 

Intervals After Enrollment 
Table 3-2 Mean Length of Enrollment, For Disenrollees and 3-63 

Those Enrolled at the End of the Calendar Year 
Table 3-3 Inpatient Utilization for Disenrollees and Enrollees 3-65 

During a 12 Month Period During Which They 

Were Enrolled in HPHC 
Table 3-4 Outpatient Utilization for Disenrollees and Enrollees 3-66 

During a 12 Month Period During Which They 

Were Enrolled in HPHC 

Table 3-5 Samples for Disenrollment Analysis 3-70 

Table 3-6 Hospital Utilization for a 3 Month Period 3-73 

Table 3-7 Part B Utilization during a 3 Month Period for Disenrollees, 3-74 

Those Always in Fee-For-Service, and Those About to 

Enroll in Managed Care 



Health Economics Research, Inc. 

crimson\final\TOC.wpt\bam 



Table of Contents: iii 



Table of Tables, Figures, and Exhibits 
(continued) 

Page 

Appendix A Profile of Harvard Pilgrim Health Care A-l 

Exhibit A-l HCHP Health Center Locations A-3 

Exhibit A-2 HCHP Medicare Plan Provider Locations A-4 

Table A-l Demographic Characteristics of Aged Medicare Enrollees: A-7 

Nationally and for HPHC Enrollees 
Exhibit A-3 HCHP Medicare Plan Provider Locations A- 10 

Appendix B 

Table B-l Sample Sizes for Access Indicators B-l 



Health Economics Research, Inc. Table of Contents: iv 

crimson\final\TOC.wpt\bam 



Executive Summary 



The purpose of this study is to develop a set of Medicare performance indicators that 
can be applied to managed care plans and to test whether these indicators can be 
implemented using elements available in a health maintenance organization data system. 
This research fits into a broader objective of developing a performance monitoring 
framework for managed care that could be used by the Health Care Financing Administration 
(HCFA), which would parallel their ongoing efforts to monitor care in the fee-for-service 
sector. The emphasis would be on monitoring across different types of Medicare 
beneficiaries, rather than monitoring the performance of any single plan. For example, 
HCFA would be interested in monitoring whether the care of patients with chronic 
conditions, such as those with diabetes, was comparable in fee-for-service and managed care. 
Similarly, the most vulnerable beneficiaries, such as the oldest-old, could be monitored to 
see whether those in managed care plans suffered relative to those in fee-for-service. This 
project serves as a pilot study for determining what measures can be constructed and 
meaningfully interpreted with "good" managed care data. 

A monitoring system may rely on several types of data: for example encounter/claims 
data, survey data, and administrative data. Doctor Colby, and Gold (1996) review alternative 



Health Economics Research, Inc. Access in Managed Care Plans: ES-1 

cri mson\final\execsumm . wpd\dpb 



Executive Summary 



sources of information for measuring performance in Medicare managed care. Our focus is 
for encounter/claims data and what can be done with them. 

The study consists of two major components. The first is the development of a series 
of Medicare performance indicators. Although in most cases, the indicators apply to both 
the managed care and fee-for-service sectors, their clinical algorithms may vary. In addition, 
some indicators apply only to managed care, given the unique features of the managed care 
encounter data. The second component of the study is to operationalize these indicators 
using Medicare fee-for-service data and data for Medicare beneficiaries enrolled in managed 
care, to determine whether the indicators can in fact be implemented and are meaningful. 

The managed care data for this analysis come from Harvard Pilgrim Health Care 
(HPHC), the largest health maintenance organization (HMO) in New England. 1 During our 
study period, HPHC contained a staff model division and an IPA-model division with very 
different data systems and incentive structures for physicians. Thus, we test indicator 
feasibility and results across the two divisions of the managed care organization as well as 
between managed care and fee-for-service. 

Defining Access and Quality in Managed Care 

Despite its importance to the national health care debate, access to care has proven 
difficult to define. Most analyses of access have relied on the framework established by 



The group was formerly known as Harvard Community Health Plan. The name was changed following a merger 
with Pilgrim Health Care in 1995. 

Health Economics Research, Inc. Access in Managed Care Plans: ES-2 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



Andersen and Aday (1978) which was developed in a context of fee-for-service medicine. 
In a fee-for-service system, the incentives are to provide more care. The major access 
concern is whether people can get into the system, or whether geographic or financial barriers 
prevent them from receiving care. As a result, access indicators have traditionally focused 
on entry into the health care system, such as number of providers available, whether patients 
have insurance, and the proportion of eligibles with at least one visit to a provider. 

In contrast to "access," the concept of "quality" has traditionally been used to 
evaluate a patient 's experience within the health care system. The distinction between access 
and quality is blurred, however, where financing and delivery systems are merged (Docteur, 
Colby and Gold, 1996). 

Thus, the discussion of access in managed care may be more about appropriateness 
of care given the incentive structure is to limit overutilization of services. Unlike the "more 
is better" attitude of fee-for-service, managed care providers act as gatekeepers to high cost 
specialty care. Thus, in managed care, access to care is not simply a matter of whether 
providers are geographically convenient, or out-of-pocket costs are affordable, but also, 
whether the gatekeeper will authorize a particular service. 

Given the difficulty of distinguishing between access measures and quality measures, 
we have decided to de-emphasize use of these terms. Instead, it may be more useful to think 
of the study as developing performance measures that can utilize claims data to determine 
whether patients receive "timely and appropriate care. " 



Health Economics Research, Inc. Access in Managed Care Plans: ES-3 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



Data and Methods 

This study uses data from three separate data systems: 

• Fee-for-service data come from HCFA's Medicare MedPAR, National 
Claims History, and Enrollment data files. These files contain the 
universe of claims for Medicare beneficiaries in fee-for-service. 

• Data for clinic encounters in the Health Centers Division (HCD), the 

staff-model division of the HMO, is kept on the Automated Medical 
Record System (AMRS), one of the earliest electronic patient record 
systems. Diagnoses, procedures, and tests are represented by COSTAR 
codes, using a system that was originally developed for the plan. (Thus, 
encounters are not coded using the ICD-9 or CPT-4 systems.) 
Information on claims and utilization outside the Centers is stored in a 
separate system using ICD-9 and CPT-4 codes. 

• Records for care in the Medical Groups Division (MGD), the IPA- 
type division of the HMO, are based on dummy claims submitted to 
HPHC to document care. These files are less rich in detail than the 
AMRS database in the HCD but are very much like most other claims 
databases, using ICD-9 and CPT-4 coding. 

Data were used for calendar years 1994 and 1995. 

To be included in the fee-for-service sample, beneficiaries were required to meet the 

following requirements: age 65 and older; having both Part A and B coverage, not enrolled 

in an HMO, and residing in the HPHC catchment area (which included much of 

Massachusetts, and southern Vermont and New Hampshire). Beneficiaries in the HPHC 

sample included those age 65 and older during the study period (January 1994-December 

1995). Beneficiaries were required to be continuously enrolled (in fee-for-service or the 

HMO) for indicator-specific periods. 



Health Economics Research, Inc. Access in Managed Care Plans: ES-4 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



Comparison of Fee-For-Service and Managed Care Results 

We constructed 19 performance indicators, grouped under five headings: primary 
care, chronic disease care, diagnosis specific care, specialty referral care, and primary care. 
Table ES-1 summarizes the results for our 19 performance indicators, which are discussed 
below. 

Preventive Care 

Given the HMO's incentives to contain costs of future care and the philosophical 
emphasis on prevention, we expected that performance in the HMO would surpass that of 
fee-for-service practice. This was clearly the case for the colon cancer screening rate, which 
was over 50 percent for both divisions of the HMO, while only 36 percent of fee-for-service 
beneficiaries received any type of screening test during the 24-month study period. 2 Nearly 
twice as high a proportion (77 percent) of aged women in the HCD received breast cancer 
screening during the 24-month period compared with women in fee-for-service (40%); 
performance in the MGD was between these two, with 67 percent of women recieving the 
test. 

We expect that much of the difference resulted from the use of an automated 
reminder system in the HCD that notifies physicians when a member is due for 
mammography. In this instance, the managed care "philosophical emphasis" on prevention 



2 

Fee-for-service coverage of fecal occult blood tests was limited during our study period, contributing to the low figure 



Health Economics Research, Inc. Access in Managed Care Plans: ES-5 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



Table ES-1 
Age- Adjusted Summary of Performance Indicators 



Preventive Care 

Breast Cancer Screening Rate 
Percentage of female beneficiaries receiving a 
mammogram during a 24 month period 

Colon Cancer Screening Rate 
Percentage of beneficiaries with a fecal occult blood test 
sigmoidoscopy, or colonoscopy during a 24 month period 

Chronic Disease Care 

Rates of Secondary Preventive Services for Diabetes Mellitus 
Percentage of beneficiaries with a diabetes diagnosis 
with each of the following during a 12 month period: 

Retinal examination 



Two or more visits with a primary care provider or 
endocrinologist 

Population-Based Admission Rate for Ambulatory Care 
Sensitive Conditions 
Admission rates per 1,000 eligibles during a 12 month period 

Rate of Pre-Hospital Care for Ambulatory Care 

Sensitive Admissions 
Percentage of beneficiaries with an ACS admission with at 
least one visit during the 60 days prior to admission 

Rate of Post-Hospital Care for Ambulatory Care 

Sensitive Admissions 

Percentage of beneficiaries with an ACS admission with at 
least one visit during the 30 days following discharge 

Anti-hypertensive Follow-up Rate 

Percentage of beneficiaries with at least one follow-up 
visit within 8 months after receiving a prescription for an 
anti-hypertensive 



HPHC 



Fee-For 
Service 



Health Medical 

Centers Groups 
Division Division 



40.8% 77.0% 64.8% 

(40.6,41.0) (75.0,79.0) (61.6,68.0) 

35.8% 58.6% 52.7% 

(35.6,36.0) (57.5,60.7) (50.5,54.9) 



54.8% 67.5% 63.9% 

(54.3,55.3) (64.7,70.3) (59.6,68.2) 

61.2% 94.6% 90.7% 

(60.7,61.7) (92.5,96.7) (90.0,91.4) 

71.9 60.1 44.4 

(71.1,72.7) (55.7,64.5) (38.4,50.4) 

80.3% 85.8% 85.3% 

(79.7,80.9) (81.7,89.9) (79.1,91.5) 



78.4% 81.8% 84.6% 

(77.8,79.9) (77.3,86.3) (78.5,90.7) 



93.0% 
(92.1,93.9) 



Health Economics Research, Inc. 

crimson\final\execsumm.wpd\dpb 



Access in Managed Care Plans: ES-6 



crimson\final\exsumtb!\ES-1\dpb 



Executive Summary 



Table ES-1 (continued) 
Age-Adjusted Summary of Performance Indicators 



HPHC 



Fee-For 
Service 



Health Medical 

Centers Groups 
Division Division 



Anti-depressant Follow-up Rate 

Percentage of beneficiaries with at least one follow-up 
visit within 8 months after receiving a prescription for an 
anti-depressant 

Diagnosis-Specific Care 

Rate of Post-Hospital Follow-up for Myocardial Infarction 
Percentage of beneficiaries hospitalized for MI 
with at least one cardiology or primary care visit within 
60 days of discharge 

Rate of Post-Hospital Follow-up for Depression 
Percentage of beneficiaries hospitalized for 
depression with at least one primary care or mental 
health visit within 14 days of discharge 

Rate of Follow-up for Abnormal Mammogram 
Percentage of female beneficiaries with an abnormal 
mammogram who receive repeat mammogram, ultrasound, 
biopsy or surgery within 1 5 days 

Specialty Referral Care 

Population-based Rate of Lens Replacement 
Rate of lens replacements per thousand beneficiaries 
during a 12 month period 

Population-based Rate of Hip and Knee Replacement 
Rate of total hip and knee replacement per thousand 
beneficiaries during a 12 month period 

Population-based Rate of Coronary Revascularization 
Rate of coronary bypass and angioplasty per thousand 
beneficiaries during a 12 month period 

Rate of Breast Cancer Oncology Follow-up 

Percentage of female beneficiaries with at least one 
oncology or general surgery visit in the 6 months 
following an initial diagnosis of breast cancer 



93.2% 
(91.7,94.7) 



73.3% 90.7% 93.2% 

(71.7,74.9) (83.8,97.6) (83.8,100) 



65.8% 64.5% 80.3% 

(62.7, 68.9) (37.3, 92.2) (57.6, 100) 



46.1% 
(34.2, 58.0) 



37.9 32.5 16.6 

(37.3,38.5) (29.1,35.9) (12.6,20.6) 



6.8 5.9 7.7 

(6.5,7.1) (4.3,7.5) (4.9,10.5) 



8.6 7.6 4.1 

(8.3,8.9) (5.7,9.5) (1.7,6.5) 



71.0% 
(63.7, 78.3) 



Health Economics Research, Inc. 

crimson\final\execsumm.wpd\dpb 



Access in Managed Care Plans: ES-7 



crimson\final\exsumtbl\ES-1 \dpb 



Executive Summary 



Table ES-1 (continued) 
Age-Adjusted Summary of Performance Indicators 



HPHC 



Fee-For 
Service 



Primary Care 

Rate of New Enrollees with a Visit 
Percentage of new enrollees with at least one visit 
during the first two months of enrollment 

Rate of Beneficiaries with a Visit 
Percentage of beneficiaries with at least one visit with a 
primary care physician or specialist during a 12 month period 

Continuity of Care Index 

Proportion of visits per patient for primary care that are with 
the patient's primary care physician 



Health Medical 
Centers Groups 
Division Division 



73.9% 48.7% 

(70.7,77.1) (45.1,52.3) 



88.4% 93.9% 90.9% 

(88.3,88.5) (91.7,96.1) (88.3,93.5) 



71.3% 
(66.7, 75.9) 



NOTE: Medicare fee-for-service did not cover routine colon cancer screening during the study period. Our rate 
may undercount the proportion of beneficiaries receiving the service if they paid out of pocket. 



Health Economics Research, Inc. 

crimson\final\execsumm.wpd\dpb 



Access in Managed Care Plans: ES-8 



crimson\final\exsumtbl\ES-1 \dpb 



Executive Summary 



and the financial incentives to provide preventive services have been institutionalized into 
a reminder system to help insure that services are in fact provided. In the MGD, which has 
no such automated system, methods of "reminding" physicians that care is due vary across 
the groups, and consequently the rate of mammography is lower. 

Chronic Disease Care 

Care for chronic diseases is an area where managed care has the potential to 
outperform fee-for-service because of the greater ability (and incentives) to coordinate care 
and manage cases through a primary caregiver. HPHC has been in the process of developing 
automated reminders for specific conditions (such as diabetes) and guidelines for treatment 
of common conditions (such as many of the ambulatory care sensitive (ACS) diagnoses). 
On the other hand, there are concerns that patients with chronic diseases, who may be quite 
expensive to treat, may be underserved and see their health deteriorate in managed care 
(Ware, et ah, 1996). The extent to which HMO initiatives to coordinate care will actually 
result in "care management" as opposed to cost reduction through "utilization management" 
has not been demonstrated. 

Our study found that both divisions of the HMO performed quite well in treating 
chronic conditions. Rates of secondary preventive services for diabetics were higher in the 
HMO than in fee-for-service, while the admission rates for ambulatory care sensitive 
conditions were lower (meaning that fewer patients reached the point which required a 



Health Economics Research, Inc. Access in Managed Care Plans: ES-9 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



hospitalization). 3 Rates of outpatient care pre- and post- ACS admission were quite high 
(80-85 percent) for both fee-for-service and managed care, indicating that most patients did 
have contact with the medical system before and after their actual hospitalization. For the 
HCD (which has computerized data on prescriptions), we also found that rates of follow-up 
for patients with prescriptions for anti-hypertensive or anti-depressant medications were quite l 
high (over 90 percent). 

Diagnosis Specific Care 

Our three indicators for diagnosis-specific care highlight the problem inherent in 
developing this type of indicator. By focusing on a very specific condition (or incident) it is 
possible to develop an indicator for which there is a consensus on appropriate treatment. 
However, the narrow focus also implies that sample sizes quickly become an issue. 

The conditions we chose (myocardial infarction, hospitalization for depression, 
abnormal mammogram) are not rare or exotic conditions among the elderly. However, given 
the number of Medicare beneficiaries enrolled in HPHC, and the resulting small samples and - 
wide confidence intervals, it is difficult to draw any conclusions regarding performance 
across the three sectors. 



Lower rates of hospitalization could result either from more timely outpatient care or from differences in the overall 
health of the populations. 



Health Economics Research, Inc. Access in Managed Care Plans: ES-10 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



Specialty Care 

Perhaps more than any other area, skeptics of managed care worry about the 
incentives to limit use of expensive specialty care. Unfortunately, provision of specialty care 
is a very difficult area to monitor, since there is so little agreement as to when referrals to i 
specialists are needed. We chose three relatively common procedures in the Medicare 
population-lens replacement, hip and knee replacement, and coronary revascularization~and 
calculated the population-based rate of each procedure. While differences in procedure rates 
may in part be attributed to differences in incidence of disease, dramatically high or low 
rates may be cause for concern. Not surprisingly, we found that the surgical rates were 
generally higher in fee-for-service than the HMO divisions. However, this may reflect 
overutilization in fee-for-service, given the incentive structure, as opposed to underutilization 
in managed care. Alternatively, both rates could be appropriate but reflect differences in 
casemix. Moreover, given the sample sizes in the HCD and MGD, the number of 
beneficiaries receiving these surgeries in the managed care setting is relatively small and 
unstable from year to year. 



Health Economics Research, Inc. Access in Managed Care Plans: ES-U 

crimson\final\execsumm wpd\dpb 



Executive Summary 



Primary Care 

The proportion of beneficiaries with at least one physician visit during a 12-month 
period is quite high for all three sectors, ranging from 88 percent in fee-for-service to 94 
percent in the HCD. A more striking comparison is found for the percentage of new 
enrollees with at least one visit during the first two months of enrollment. This rate is much '- 
higher for the HCD than the MGD, and the gap narrows, but does not disappear as the time 
horizon is expanded. The HCD's high rate reflects its aggressive campaign to triage and 
assess high risk patients. The lower rate for the MGD may reflect movement of patients into 
the MGD who join HPHC from another HMO or fee-for-service but do not change 
physicians. These patients would not be assessed as new patients, since they continue to visit 
the same medical group and physician as before joining HPHC. 

Implications for Developing a Monitoring System 

This project was intended to serve as a pilot study for determining what measures 
could be constructed~and meaningfully interpreted-with "good" managed care data. It was - 
designed to help HCFA in the development of a framework for monitoring managed care. 
This would parallel their ongoing efforts to monitor care in the fee-for-service sector. Hence, 
we conclude with a discussion of "lessons learned" during the course of the study that 
addresses the implications for applying a set of performance measures to other health plans 
or providers. 



Health Economics Research, Inc. Access in Managed Care Plans: ES-12 

crimson\final\execsumm wpd\dpb 



Executive Summary 



Constructing the Indicators. Once we had developed the final set of indicators, 
they were constructed using the different claims/encounter databases for Medicare fee-for- 
service, the HCD, and the MGD. In this section, we briefly describe some of the difficulties 
encountered in developing and interpreting the indicators. 

Reconciling Differences in Coding Systems. The fee-for-service and MGD data, '" 
along with the HCD institutional data, used ICD-9 diagnosis and CPT-4 coding. The HCD 
ambulatory claims used the COSTAR coding system that was originally developed by 
Harvard Community Health Plan. 

Because of the different coding schemes, we were forced to develop comparable 
definitions for identifying diagnoses and procedures for all indicators based on outpatient 
care. In defining the indicators, two questions were considered: 

• Is there an identical (or similar) code in each system? 

• Are physicians equally likely to use the code (given a procedure was 
performed or condition was observed) in each system? 

For many indicators, developing similar definitions was quite straightforward, as 

COSTAR coding corresponded quite closely to ICD-9 or CPT-4 coding. For example, the - 
list of codes for colorectal cancer screening tests is fairly extensive, but the definitions of 
codes correspond closely in ICD-9 and COSTAR coding. 

The most difficult definition to develop was for retinal screening for diabetics. The 
COSTAR system has codes for eye examinations. However, given the payment structure of 
the HCD, optometrists/ophthalmologists have no incentive to code that a specific test was 



Health Economics Research, Inc. Access in Managed Care Plans: ES-13 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



performed; rather, they are more likely to code the findings of the test. We found that they 
often coded a diagnosis that would normally require a retinal exam without coding the exam 
itself. Thus, rather than selecting a few COSTAR codes that would correspond to the CPT 
codes for retinal exam, we were forced to rely on a series of diagnostic codes that could only 
be found if a retinal exam were performed. If a physician failed to code the exam, and found '* 
no abnormalities, we may underestimate the numerator for this indicator. 

In fee-for-service, physicians may bill for a visit rather than an eye exam, since 
payment may differ for the two codes. If this happened, we also may undercount in fee-for- 



service. 



A second coding issue is the appearance of "rule out" diagnoses in the data. The 
HCD data system allows physicians to mark a diagnosis as being a "rule out"-although it 
is not clear that these are always indicated. The fee-for-service and MGD data have no such 
marker for "rule out" diagnoses, and it is impossible to determine which are intended as 
definitive diagnoses and which are coded as "rule outs." For illnesses which are likely to 
have a high proportion of "rule out" diagnoses in the claims, this difference in coding 
complicates development of similar samples. For the diabetes indicators, we required that 
the diagnosis be attached to a physician claim (rather than, say, a laboratory claim) in an 
attempt to reduce the number of "rule outs." Given the significant number of beneficiaries 



Health Economics Research, Inc. Access in Managed Care Plans: ES-14 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



in all three data sets with only one diabetes diagnosis, any attempt to identify all patients with 
the disease is likely to either miss some true cases or include some rule-out diagnoses. 4 

Variations in Data Set Structure. In addition to differences in data coding systems, 
the structures for the data sets varied across the three settings. For example, all of the data 
systems we worked with had separate files for inpatient institutional claims. However, the t 
actual claims stored in the hospital file differed across the data systems. Initial attempts to 
locate mammogram codes for the MGD identified only 2 percent of women with claims for 
a mammogram during a two-year period, including no claims in 1994. Further investigation 
revealed that claims for Medicare recipients were not located in the ambulatory claims files, 
but in hospital claims files. In contrast, in fee-for-service data, mammography claims can 
be found in the physician/supplier file, the outpatient department file, or both files. 

This example highlights one danger of working with unfamiliar data sets. If all data 
(or virtually all data) are missing, as was the case with mammography in the MGD, it is easy 
to recognize the problem. If some of the data are missing, as was the case in the fee-for- 
service physician/supplier file, it can be much more difficult to recognize that the problem * 
exists. 



4 

HEDIS attempts to eliminate "rule-out" diagnoses by requiring that the diagnosis appear twice during the calendar 
year. The disadvantage of this approach is that it may bias estimates of performance indicators upwards, if some 
patients have only one diagnosis because they are low utilizers of care. 



Health Economics Research, Inc. Access in Managed Care Plans: ES-15 

crimson\final\execsumm.wpd\dpb 



Executive Summa 



£L_ 



Costs of Processing Data. The cost of processing claims can be high, especially 
when it is necessary to search through a large database multiple times, for example, to first 
search an outpatient database to identify all claims with a particular diagnosis, and then 
search again to pull all claims for beneficiaries with that diagnosis. 

For a medical record database, such as HPHC's Automated Medical Record System, "" 
the cost can be prohibitive, even on relatively small samples of data. Since the data source 
is a medical record, rather than a claim, data processing of relatively small samples of data 
becomes time-consuming and expensive. Thus, in estimating the burden on plans from 
implementing a monitoring system, the data processing requirements should not assume that 
all plans have access to claims data and can process data in a similar manner. 

Limitations in Sample Sizes. One of our criteria for selecting indicators was that 
they be related to a high-incidence disease or a high-incidence procedure. Given the limited 
number of indicators that can be monitored, we did not want to select a rare condition (or 
procedure) upon which to base a performance measure. Even using relatively common 
diseases and procedures, our samples were quite small for several indicators in the HCD and - 
the MGD, which had roughly 11,000 and 5,500 aged Medicare members, respectively. 
Sample size decreases even more for indicators that require a lengthy continuous enrollment 
period. Even where overall samples were relatively large, we were often limited in the 
stratifications that could be made. 



Health Economics Research, Inc. Access in Managed Care Plans: ES-16 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



We developed all indicators and presented rates and confidence intervals regardless 
of sample size. (Obviously, the likelihood of detecting statistically meaningful differences 
is much lower for the indicators based on very small samples.) Given the exploratory nature 
of this project, we felt this was an appropriate approach. 

For a set of performance indicators intended as a "report card," an approach that does *• 
not rely on audience familiarity with confidence intervals and statistical tests may be more 
appropriate. For example, HEDIS 3.0 specifies that if a measure applies to fewer than 100 
members, the plan should report a 95 percent confidence interval, and that measures based 
on fewer than 100 members should not be used for comparisons among health plans. 
Moreover, HEDIS specifies that measures should not be reported when there are fewer than 
30 members in the denominator. Our post-depression follow-up measure would not have 
been reported using this criteria, and samples for the myocardial infarction and abnormal 
mammogram follow-ups both fell below the 100 member threshold. 

Interpreting the Results. Claims-based monitoring systems can tell us what 
occurred in a patient's medical care, but not why. For example, the results of our data 
processing indicated that the rate of mammography was much higher in the HCD than in the 
MGD or fee-for-service. However, the claims cannot give us information on whether the 
difference resulted from provider willingness to encourage mammography, patient 
willingness to have the procedure, availability of convenient locations/hours for 
mammography services, or some other reason. In fact, we believe the difference is largely 



Health Economics Research, Inc. Access in Managed Care Plans: ES-17 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



attributable to the HCD automatic reminder system, that prompts physicians when a 
beneficiary is due to receive a mammogram. 

The advantage of the claims-based system is that it can, at relatively low cost, flag 
areas where the system is doing well or poorly. This allows policy-makers to concentrate 
further effort on areas where improvements are needed. By combining a claims-based 
system with other approaches to gauging access and quality, such as surveys and chart audits, 
we can gain a much more complete picture of plan performance. 

Conclusion 

Genera lizability of our Experience. The purpose of this study was to develop a set 
of Medicare performance indicators that could be applied to managed care plans and to test 
whether these indicators could be implemented using elements available in an HMO data 
system. This project was intended to serve as a pilot study for determining what measures 
can be constructed, and meaningfully interpreted, with "good" managed care data. 

We began the study knowing that our HMO data were of higher quality than that 
found in many managed care organizations. Numerous studies have been published using 
diagnosis and procedure data from the HCD's Automated Medical Record System (studying 
conditions as diverse as streptococcal pharyngitis, hypertension, and bipolar disorder). Data 
from the MGD have not been used for published research to the same extent as data from the 
HCD. However, the plan has used the data bases for its own internal analysis. Thus, 



Health Economics Research, Inc. Access in Managed Care Plans: ES-18 

crimson\final\execsumm.wpd\dpb 



Executive Summary 



although we have constructed a set of performance indicators with two types of HMO data, 
it is not clear whether the data systems of other managed care organizations can support the 
same types of analysis. Many pressures (including HEDIS) are pushing managed care 
organizations to improve their data systems. Thus, construction of performance indicators 
is much more feasible than it would have been even a few years ago. 

Next Steps. For this project, we developed a set of 19 performance indicators, 
several of which were constructed using alternate methodologies (for example, varying the 
episode length). While we constructed multiple rates in order to test the sensitivity of our 
results to varying definitions, it would be desirable to determine the preferred definition that 
would be reported as part of the performance monitoring system. 

More importantly, it would be desirable to replicate this study using data from other 
health plans. Using data from two divisions of HPHC, we have found that our indicators can 
be constructed, and comparisons among the two divisions and fee-for-service practice show 
meaningful differences in the performance of the three sectors. We have also found, 
however, that differences in databases can complicate construction and interpretation of the 
indicators. Extending the work to include data from other health plans would be the next 
step towards developing these indicators into a monitoring system for managed care 
performance. 



Health Economics Research, Inc. Access in Managed Care Plans: ES-19 

crimson\final\execsumm.wpd\dpb 



1 



Introduction 



1.1 Objectives of the Project 

Both the President and Congress have proposed significant changes in Medicare, 
including fundamentally restructuring the way care is organized and delivered as well as 
generating substantial reductions in the growth of expenditures. These proposed changes in 
the health care delivery system are primarily being driven by cost, accompanied by an 
emphasis on fostering competition, "managing care, " creating networks of "preferred " 
providers, and assigning "gatekeeper" physicians as conduits to services. A common 
element of many of these initiatives is the realization that Americans will no longer enjoy 
unquestioned, unfettered access to whatever specific services they desire or that their 
physicians are motivated to recommend. 

These anticipated changes heighten the need for long-term, continuous, monitoring 
of the care received by Medicare program beneficiaries. The Health Care Financing 
Administration (HCFA) has a long history of monitoring access to care for Medicare 
beneficiaries, but most of these efforts have focused on the fee-for-service sector. 1 Given the 
cost-containment incentives that providers face in managed care programs, and the growing 



The historical focus on fee-for-service resulted from the overwhelming majority of beneficiaries belonging to this sector 
and the availability of claims data for fee-for-service care. For managed care there have always been more extensive up- 
front requirements (relating to who can be a contractor) and ongoing monitoring through site visits, PROs, etc. 

Health Economics Research, Inc. Access in Managed Care Plans: 1-1 

crimson\final\chap 1 . wpd\dpb 



Cha P ter I Introduction 

significance of these programs for Medicare beneficiaries, efforts to monitor access must be 
broadened to include beneficiaries in managed care. 

Measuring access to care for managed care enrollees is more difficult than simply 
taking indicators that have been developed using fee-for-service data and applying them to 
managed care plans. Little information on services provided to patients has historically been * 
available from most managed care plans, although the situation is changing rapidly. One of 
the administrative advantages to capitated payment systems is the absence of the need for 
claims. Services are provided, but no bills are submitted to the payer (e.g., Medicare). While 
many managed care plans do maintain encounter data for their own internal management and 
quality assurance purposes, these data vary markedly in their completeness, reliability, and 
availability to researchers outside the managed care organization. Furthermore, there is 
currently no standard method of collecting and reporting such encounter data across plans. 

The purpose of this study is to develop a set of Medicare performance indicators that 
can be applied to managed care plans and to test whether these indicators can be 
implemented using elements available in a health maintenance organization (HMO) data - 
system. This research fits into a broader objective of developing a performance monitoring 
framework for managed care that could be used by HCFA, which would parallel their 
ongoing efforts to monitor care in the fee-for-service sector. The emphasis would be on 
monitoring across different types of Medicare beneficiaries, rather than monitoring the 
performance of any single plan. This project serves as a pilot study for determining what 
measures can be constructed and meaningfully interpreted with "good " managed care data. 



Health Economics Research, Inc. Access in Managed Care Plans: 1-2 

crimson\final\chap 1 . wpd\dpb 



Cha P terl Introduction 

A monitoring system may rely on several types of data; for example, 
encounter/claims data, survey data, and administrative data. Docteur, Colby, and Gold 
(1996) review alternative sources of information for measuring performance in Medicare 
managed care. Our focus is on encounter/claims data and what can be done with them. 

The study consists of two major components. The first is the development of a series >■ 
of Medicare performance indicators. Although in most cases, the indicators apply to both 
the managed care and fee-for-service sectors, their clinical algorithms may vary. In addition, 
some indicators apply only to managed care, given the unique features of the managed care 
encounter data. The second component of the study is to operationalize these indicators 
using Medicare fee-for-service data and data for Medicare beneficiaries enrolled in managed 
care, to determine whether the indicators can in fact be implemented and are meaningful. 

The managed care data for this analysis come from Harvard Pilgrim Health Care 
(HPHC), the largest HMO in New England 2 . During our study period, HPHC contained a 
staff model division and an IPA-model division with very different data systems and 
incentive structures for physicians. Thus, we can test indicator feasibility and results across 
the two divisions of the managed care organization as well as between managed care and fee- 
for-service. Appendix A contains material describing the structure and systems of the two 
HMO divisions. This helps to interpret differences in performance between the two 
divisions, and between the HMO and fee-for-service. 



2 

The group was formerly known as Harvard Community Health Plan. The name was changed following a merger with 
Pilgrim Health Care in 1995. 

Health Economics Research, Inc. Access in Managed Care Plans: 1-3 

crimson\finaI\chap 1 . wpd\dpb 



Cha P terl Introduction 

1.2 Monitoring Performance in Managed Care and Fee-for-Service 

Numerous studies give reason to suspect that performance may differ between 
managed care and fee-for-service settings. Under fee-for-service reimbursement, providers 
receive additional payment for each billable service provided to the patient. As a result, their 
financial incentive is to provide more services (and submit more claims) to increase revenue. 
In contrast, under a managed care risk-contract, the capitated reimbursement is fixed 
regardless of the services provided. Thus, the financial incentive under the contract is to 
limit use of expensive resources, particularly if their ability to improve health or reduce 
future expenses is ambiguous. Moreover, there is no direct incentive in a capitated system 
to maintain complete claims data for each patient encounter. 

What differences between managed care and fee-for-service might we expect given 
the different incentive structures? First, managed care providers may be more likely than 
fee-for-service providers to provide preventive care (immunizations) or screening services 
(mammography, check-ups) that may reduce future costs of treatment by allowing early 
treatment. Bernstein et al. (1991) support this hypothesis, having found that HMOs had - 
higher rates of preventive services even when compared to fee-for-service plans that had no 
out-of-pocket payments. Riley et al. (1994) found that HMO enrollees were diagnosed 
earlier than fee-for-service enrollees for cancers of the female breast, cervix, colon, and 
melanomas, although they were diagnosed at a later stage for stomach cancer. 

However, the cost-containment incentives of managed care may result in underservice 
and suboptimal care, particularly for some types of conditions (Spitz, 1979; Rowland and 



Health Economics Research, Inc. Access in Managed Care Plans: 1-4 

crimson\final\chap 1 wpd\dpb 



Chapter 1 Introduction 

Lyons, 1987). Managed care providers may be less likely to offer access to expensive 
technology (MRI or CT), expensive procedures (bypass surgery), or access to specialty care 
(cataract surgery) (Goldzweig, et al. 1997). Vulnerable subgroups -- the oldest old, those 
with functional impairments, and those in poorer health -- may be particularly affected by 
incentives to limit resource use (Nelson, et al. 1997; Ware et al. 1996). Empirical research * 
points to lower per-patient expenditures among recipients of pre-paid care than among 
comparable patients with fee-for-service insurance (Manning, et al. 1984; Greenfield et al. 
1992; Miller and Luft, 1994). 3 

Given the concerns that managed care may be "underperforming " relative to fee-for- 
service on some measures, it is important to benchmark managed care performance against 
that in the fee-for-service sector. Otherwise, managed care performance may be compared 
against some "ideal " performance that is not being achieved elsewhere. Of course, even 
when we benchmark we cannot always distinguish underperformance in one sector from over 
performance in another. 

1.3 Defining Access in a Managed Care Environment 

Despite its importance to the national health care debate, access to care has proven 
difficult to define. Most analyses of access have relied on the framework established by 
Andersen and Aday (1978) which was developed in a context of fee-for-service medicine. 
In a fee-for-service system, the incentives are to provide more care. The major access 



However, Brown et al. (1993) found that outcomes are comparable for HMO and fee-for-service patients, suggesting that 
the lower level of services appears to be due to the elimination of discretionary services. 

Health Economics Research, Inc. Access in Managed Care Plans: 1-5 

crimson\final\chapl wpd\dpb 



Chapter 1 Introduction 

concern is whether people can get into the system, or whether geographic or financial barriers 
prevent them from receiving care. As a result, access indicators have traditionally focused 
on entry into the health care system, such as number of providers available, whether patients 
have insurance, and the proportion of eligibles with at least one visit to a provider. 

In contrast to "access," the concept of "quality" has traditionally been used to fc 
evaluate a patient 's experience within the health care system. The distinction between access 
and quality is blurred, however, where financing and delivery systems are merged (Docteur, 
Colby and Gold, 1996). 

Thus, the discussion of access in managed care may be more about appropriateness 
of care given the incentive structure is to limit overutilization of services. Unlike the "more 
is better" attitude of fee-for-service, managed care providers act as gatekeepers to high cost 
specialty care. Thus, in managed care, access to care is not simply a matter of whether 
providers are geographically convenient, or out-of-pocket costs are affordable, but also, 
whether the gatekeeper will authorize a particular service. 

Given the difficulty of distinguishing between access measures and quality measures, 
we have decided to de-emphasize use of these terms. Instead, it may be more useful to think 
of the study as developing performance measures that can utilize claims data to determine 
whether patients receive "timely and appropriate care. " 



Health Economics Research, Inc. Access in Managed Care Plans: 1-6 

crimson\final\chap 1 . wpd\dpb 



Chapter 1 Introduction 

1.4 Conceptual Framework 

We use a three dimensional model to capture aspects of performance. This 
conceptual framework is shown in Exhibit 1-1. The three dimensions of performance 
measurement in our framework are: resource availability, utilization, and satisfaction. 
Resource availability measures reflect the availability (within the network) and convenience l 
(location, hours) of providers and services. These indicators measure "potential access " for 
patients. 

Within the centerpiece of utilization measures, we have five subcomponents for types 
of care: (1) preventive care, (2) chronic disease care, (3) diagnosis-specific care, (4) specialty 
referral care, and (5) primary care. These five subcomponents capture aspects of care for 
which managed care plans have very different financial incentives than do fee-for-service 
providers. 

• Preventive care includes immunizations and screening tests. This is an area of 
care in which HMOs may surpass fee-for-service practice, given the HMO 
incentives to contain costs of future care and their philosophical emphasis on 
prevention. 

• Chronic disease care measures examine whether patients are receiving 
appropriate follow-up care for selected chronic conditions. The financial 
incentives for capitation again may lead HMOs to provide superior care to 
prevent future complications, although these incentives will vary by condition 
and type of intervention. 

• Diagnosis specific care examines treatment for acute conditions or episodes. 
Again, we would expect HMOs to provide care that would prevent complications 
and expenses in the future. 

• Specialty referral care is an area of concern for HMO enrollees. Given financial 
incentives to reduce costs, patients may not be receiving specially referrals on a 
timely or appropriate basis, or may not be receiving costly services/procedures 



Health Economics Research, Inc. Access in Managed Care Plans: 1-7 

crimson\final\chap 1 . wpd\dpb 



Chapter 1 



Introduction 



from specialty providers. Defining appropriate specialty care is particularly 
problematic, given the lack of consensus on when it should be sought. 

• Finally, we have a broad category for primary care. This includes broader 
measures of whether patients "get into the system " for any care. 

EXHIBIT 1-1 

Monitoring Delivery of Timely and Appropriate Care: 
Conceptual Framework 



Resource 
Availability 



Utilization 



Satisfaction 




Primary Care 




Given that our study is testing measures that can be developed using claims/encounter 
data, our indicators focus on the utilization portion of performance measurement. Although 
our case study investigates issues of plan structure and satisfaction measurement in a well- 
established managed care organization, our indicators reflect the various aspects of 
utilization we have described above. 



Health Economics Research, Inc. 

crimson\final\chap 1 . wpd\dpb 



Access in Managed Care Plans: 1-8 



Cha P ter 1 Introduction 

1.5 Selection of Indicators 

Crucial to the monitoring effort is the selection of appropriate and measurable 
indicators. Selection of indicators should be based on their policy relevance, the availability 
of data, and the extent to which various measures address important public health priorities. 
Absolutely essential is that the indicator can be constructed with available data. In addition, l 
each indicator should meet at least one of the following criteria: 

• Be of epidemiological or clinical importance; 

• Have sufficient clinical consensus on its need or associated treatment 
protocol 

• Be a high incidence procedure or related to a high incidence disease; 

• Have a high expected health impact; or 

• Be related to costly services. 

A natural inclination in comparing performance measures for managed care and fee- 
for-service medicine is to use fee-for-service as a benchmark of "appropriate provision of - 
services, " and assume that lower levels of use for managed care patients represents "poorer 
performance." However, for many conventional measures of performance, there is little or 
no evidence that fee-for-service represents some optimal standard of care. For instance, a 
finding that managed care patients average fewer visits per year than fee-for-service patients 
does not necessarily mean that managed care patients have too few visits. An alternative 
interpretation would be that fee-for-service patients are overutilizing care, or that differences 



Health Economics Research, Inc. Access in Managed Care Plans: 1-9 

crimson\final\chap 1 . wpd\dpb 



s. 



Cha P ter I , Introduction 

in patient health status account for the differential. Thus, to the extent possible, performance 
indicators should be based on clinical standards of care, supported (directly) by published 
research and (indirectly) by guidelines that are evidence-based, and the proportion of patients 
for whom these standards are met, rather than on vague measures of usage (such as average 
number of visits per beneficiary) that are difficult or impossible to interpret. 

There may also be trade-offs between ease/accuracy of measurement and salience in 
selection of indicators. For example, screenings and immunizations provide easy to interpret 
indicators and meet several of the selection criteria. The mammography rate for women is 
a well-established performance indicator (with a strong clinical consensus and high expected 
impact on health status). In addition, the results are easily interpretable: higher rates of 
mammography are better than lower rates. Although other screening tests may fit all these 
criteria, a monitoring system must have broader focus. 

Those wary of managed care organizations are often concerned that they will under- 
provide high-cost procedures and treatments. However, compared to screenings and 
immunizations, utilization rates of specialty care are difficult to interpret. Without detailed 
clinical data, it is difficult to evaluate a rate of lens replacement, since the appropriate rate 
of surgery will differ across different populations. Lower rates in managed care could 
indicate underutilization, or, given the incentive structure of the fee-for-service system, may 
imply overutilization in fee-for-service. Nonetheless, it is important to address the area of 
specialty care, even if the performance indicators may be less well supported by clinical 
consensus than an indicator like mammography. 



Health Economics Research, Inc. Access in Managed Care Plans: 1-10 

crimson\final\chap 1 wpd\dpb 



Chapter 1 Introduction 

1.6 Indicators Included in This Study 

To develop indicators of performance, we began with the framework of Siu et al. 
(1992), of examining leading causes of morbidity and mortality among the elderly. As 
discussed above, our goal was to generate a list of indicators meeting the selection criteria, 
that were spread across the five different subcomponents of utilization in our framework. 
The final list of indicators is presented in Table l-l. 4 Most indicators were developed for all 
three sectors--fee-for-service, Health Centers Division (HCD) within HPHC (the staff model 
division), and the Medical Groups Division (MGD) of HPHC (the EPA division). However, 
a few took advantage of unique aspects of data available in the HCD and were constructed 
only for that division. 

• The two preventive care indicators, breast cancer screening rate and colon 
cancer screening rate were developed for both fee-for-service and 
managed care. 5 

• The chronic disease indicators include two measures for diabetic care, 
rate of retinal eye exam and proportion of patients with at least two visits 
during a twelve-month period. Chronic disease care also includes 
treatment patterns for ambulatory care sensitive (ACS) conditions, 
including admission rates during a twelve month period, and rates of care 
prior to admission and following discharge from the ACS hospitalization. 
The chronic condition indicators also take advantage of the HCD data 
base that provides prescription drug information, to calculate rates of 
follow-up visits for patients on antihypertensives or antidepressants. 



Section 4.2 describes a series of other indicators that were considered for construction, but were at some point eliminated 
from the list. 

Detailed definitions of indicators and discussion of their construction can be found in Chapter 3. 

Health Economics Research, Inc. Access in Managed Care Plans: 1-11 

crimson\final\chap 1 . wpd\dpb 



Chapter 1 



Introduction 



Table 1-1 
Summary of Medicare Performance Indicators 





Fee-for 


Harvard Pilgri 


m Health Care 




Health Centers 


Medical Groups 


Indicator 


Service 


Division 


Division 


Preventive Care 








Breast cancer screening rate 


X 


X 


X 


Colon cancer screening rate 


X 


X 


X 


Chronic Disease Care 








Diabetes: 








Retinal examination rate 


X 


X 


X 


Visit rate 


X 


X 


X 


Ambulatory care sensitive conditions: 








Admission rate 


X 


X 


X 


Rate of pre-hospital care 


X 


X 


X 


Rate of post-hospital care 


X 


X 


X 


Anti-hypertensive follow-up rate 


- 


X 


— 


Anti-depressant follow-up rate 


- 


X 


- 


Diagnosis-Specific Care 








Rate of post-hospital follow-up for: 








Myocardial infarction 


X 


X 


X 


Depression 


X 


X 


X 


Rate of follow-up for abnormal 


- 


X 


— 


mammogram 








Specialty Referral Care 








Population based rate of 


X 


X 


X 


lens replacement 








Population based rate of hip and knee 


X 


X 


X 


replacement 








Population based rate of coronary 


X 


X 


X 


revascularization 








Rate of breast cancer oncology follow-up 


- 


X 


- 


Primarv Care 








New enrollee visit rate 


- 


X 


X 


Annual visit rate 


X 


X 


X 


Continuity of care index 


- 


X 


- 



NOTE: 

X Indicator constructed for this setting. 
- Indicator not constructed for this setting. 



Health Economics Research, Inc. 

crimson\finaI\chap 1 wpd\dpb 



Access in Managed Care Plans: 1-12 



Chapter 1 Introduction 



• Diagnosis-specific indicators include rates of post-hospitalization follow- 
up for individuals with myocardial infarction or depression (calculated for 
both fee-for-service and managed care) and rate of follow-up for 
abnormal mammogram, which again takes advantage of a unique aspect 
oftheHCDdata. 

• Specialty care indicators include rate of oncology or surgery follow-up for 
breast cancer patients, and population-based rates for three common 
procedures among the elderly, lens replacement, major joint replacement, 
and coronary revascularization. 

• The primary care indicators include a continuity of care index, a new 
enrollee visit rate, and the proportion of patients receiving an annual visit. 



1.7 Analysis of Disenrollees 

Generating performance indicators is a time- and resource-consuming activity for any 
managed care organization. Thus, any set of performance indicators that an HMO might be 
required to report must be relatively limited, and cannot cover all aspects of care. Although 
our list of indicators was constructed to cover different types of care (e.g., preventive, 
specialty) and include different types of conditions (e.g., diabetes, mental health, myocardial 
infarction), they obviously cannot cover the entire spectrum of care. It is conceivable, for 
example, that an HMO might excel at treating diabetics, but have poor management of other 
chronic conditions. 

Disenrollment rates are often considered another sentinel indicator of HMO 
performance. Among HMOs, a high disenrollment rate may signal poor performance, as 
members leave for either fee-for-service or another HMO. High disenrollment rates for some 
group of members (for example, the oldest old, those with chronic conditions) may signal 



Health Economics Research, Inc. Access in Managed Care Plans: 1-13 

crimson\fmal\chap 1 . wpd\dpb 



Chapter 1 Introduc tion 

dissatisfaction with the plan's performance in providing care. Thus, although the 
disenrollment analysis is in many ways less "precise " than the utilization-based performance 
indicators, it may detect other areas in which patients are satisfied or dissatisfied with their 
care. 

1.8 Organization of the Report 

The remainder of the report is divided into three chapters. Chapter 2 describes the 
data sources used in the analysis, our sample selection criteria, and an overview of methods, 
such as age adjustment, that apply to all indicators. Chapter 3 presents detailed definitions 
for each of the indicators, the process of constructing the indicators, and results for the three 
sectors. Because of our emphasis on determining whether indicators can be constructed in 
a meaningful manner, we report indicator-specific methodological and data issues in this 
chapter, rather than in Chapter 2 which provides a broader overview. Chapter 4 discusses 
these results and the implications of our study for developing a monitoring system. 

We also provide three appendices to the report. Appendix A presents a discussion . 
of the HMO which provided data for the analysis, which serves to explain differences in the 
performance of the two divisions, and to address the resource availability and satisfaction 
aspects of our conceptual framework. Appendix B presents detailed sample sizes for each 
of the indicators. Appendix C includes file layouts for the HPHC data sets that were used 
in the analysis. 



Health Economics Research, Inc. Access in Managed Care Plans: 1-14 

crimson\finaI\chapl wpd\dpb 



2 



Data Sources and 
Methods Overview 



This chapter describes data sources and methodology used to construct the 
performance indicators. We first describe in detail the nature of the data files from HPHC 
and HCFA that were used to construct the indicators. Next we describe our sample selection 
and method of defining the catchment area from which fee-for-service beneficiaries were 
selected. Then we discuss several technical issues we encountered in constructing the 
indicators and the types of external benchmarks we used for comparison with our results. 

2.1 Data Sources 

Our managed care data come from two divisions of Harvard Pilgrim Health Care: the 
staff model Health Centers Division and the IP A model Medical Groups Division. The two 
divisions have separate and very different data systems. Fee-for-service data come from 
HCFA's Medicare records. 

2.1.1 Harvard Pilgrim Health Care (HPHC) 

Data for constructing the access indicators for beneficiaries enrolled in HPHC come 
from the following sources: 1 



Appendix C provides a more detailed description of data elements available in the HPHC systems. 

Health Economics Research, Inc. Access in Managed Care Plans: 2-1 

crimson\final\chap2.wpd\dpb 



Chapter 2 Methods 

Enrollment Data : The Membership Utility Program (MUP) is a SAS file created from 
enrollment information which covers all divisions of HPHC and includes information about 
insurance coverage. It also includes demographic information such as date of birth, gender, 
and zip code of residence. 

Health Centers Division (HCDY The HCD is the staff model division of HPHC. * 
Information on clinic encounters in the 14 sites is kept electronically. The system, called the 
Automated Medical Record System (AMRS), was developed specifically for HPHC and was 
one of the earliest electronic patient records. Diagnoses, procedures, and tests are 
represented on the file by COSTAR codes, an ambulatory medical record system that was 
originally developed to support HPHC medical practices. 2 

The HCD uses a computer system called TOPPS for handling all claims and 
utilization "outside" the Centers themselves. This system uses ICD-9 codes for claims from 
hospitals. The structure contains up to six diagnoses and up to 3 surgical procedures. 

Medical Groups Division (MGD) : In the MGD, HPHC contracts with groups of 
physicians who are geographically dispersed throughout the region and are not on HPHC's 
staff. Records for care in the MGD are based on dummy claims submitted to HPHC to 
document care. These files are less rich in detail than the AMRS database in the HCD but 
are very much like most other claims databases. The Clinic File contains claims for services 
provided in the offices of the primary care providers in the MGD, the Outpatient File 
contains claims for outpatient services not provided by the primary care provider, i.e. 



Thus, data from the clinics do not use ICD-9 or CPT-4 coding. 



Health Economics Research, Inc. Access in Managed Care Plans: 2-2 

crimson\final\chap2.wpd\dpb 



Chapter 2 Methods 

referrals, and the Institutional File contains claims for hospitalizations. These files contain 
diagnosis and procedure information using ICD-9 and CPT-4 codes. The MGD institutional 
file differs from the HCD file in two main ways: (1) the MGD file contains up to 3 diagnoses 
(versus 6 in HCD), and (2) the MGD file contains up to 6 surgical procedures (versus 3 in 
HCD). 

2.1.2 Fee-for-Service 

Data for constructing the indicators for Medicare beneficiaries come from the 
following sources: 

Enrollment Data : The Denominator file contains information on all Medicare 
beneficiaries. Variables on the file include zip code of residence, reason for eligibility, 
whether the individual receives Part A and/or Part B benefits (with a monthly indicator), and 
whether the individual belongs to an HMO (with a monthly indicator). The cross-reference 
file contains information on beneficiaries whose HICNOs (identifying numbers) change, 
allowing these beneficiaries to be tracked throughout the study period. Use of the cross- - 
reference file is especially important for indicators that requiring tracking the same individual 
across a longer time period (as the likelihood of the identifier changing increases over time) 
and for female beneficiaries (who are more likely to receive benefits through a spouse's work 
history, and whose HICNO will change with changing marital status). 3 The cross-reference 
file was used in the construction of all indicators. 



For example, our rate of mammography using physician/supplier claims during a 24 month period for the four-state 
New England area rose from 34.9 percent without cross-referencing to 38. 1 percent after using the cross-reference file. 

Health Economics Research, Inc. Access in Managed Care Plans: 2-3 

crimson\fi nal\chap2 , wpd\dpb 



Cha P ter2 Methods 

Hospitalization Data: The MedPAR file contains information on all hospitalizations 
for Medicare beneficiaries. Variables on the file include patient's HICNO, date of admission 
and discharge, up to 10 diagnosis and 10 procedure codes, and patient's DRG. 

Physician Utilization Data: The Part B physician/supplier files contain the universe 
of physician claims for beneficiaries in our catchment area. Variables on the file include «■ 
patient's HICNO, date of service, physician specialty, a unique physician identifier (UPIN), 
and diagnosis and procedure codes. 

Hospital Outpatient D epartment Data : The hospital outpatient file contains a 100 
percent sample of claim-level information on procedures performed in these facilities. 
Variables on the file include patient's HICNO, date of service, and procedure codes. 

2.2 Sample Selection 

The sample criteria vary considerably across our indicators, based on the relevant 
population and time frame. This section describes our overall criteria for beneficiaries to be 
included in our analysis. In Chapter 3 we describe in detail the criteria for each indicator. - 
These were constructed to require continuous enrollment across the analytic period (e.g. to 
allow 60 day follow-up) 4 . 



4 

Alternatively, we could have allowed those eligible for a portion of the period to enter the analysis, and weighted the 
indicators by the fraction of the period for which they were eligible. The approach we took, requiring continuous 
eligibility across the entire period, is consistent with the approach HCFA uses in its calculation of mammography rates. 
It is similar to that used in HEDIS, which requires continuous enrollment but allows one short break in enrollment. 

Health Economics Research, Inc. Access in Managed Care Plans: 2-4 

crimson\final\chap2.wpd\dpb 



Chapter 2 Methods 

2.2.1 HPHC 

For both divisions of the health plan, Medicare beneficiaries who were age 65 and 
older during the study period (January 1 994-December 1995) were identified. The HCD 
contained roughly 1 1,000 Medicare members, while the MGD had about half that number. 
Members who switched from one division to the other during the study period were not 
included in the sample. Only 2 percent of the sample was lost due to this restriction. 
Members were required to bo continuously enrolled in the plan for a members period defined 
on an indicator - specific basis. 

2.2.2 Fee-For-Service 

To be included in the fee-for-service sample, beneficiaries were required to meet the 
following requirements: age 65 and older as of January 1, 1994; having both Part A and B 
coverage; and not enrolled in an HMO. Beneficiaries meeting these criteria were identified 
from the denominator file. In addition, the beneficiary's place of residence, as indicated on 
the denominator file, was required to be within the HPHC catchment area, as described 
below. 

2.2.3 Defining the HPHC Catchment Area 

HPHC constructed a list of zip codes in which their elderly beneficiaries resided 
during the study period, and the number of beneficiaries living in each zip code. Based on 
this listing, we acquired claims data for beneficiaries living in four states: Massachusetts, 



Health Economics Research, Inc. Access in Managed Care Plans: 2-5 

crimson\final\chap2.wpd\dpb 



Chapter 2 Methods 

New Hampshire, Rhode Island, and Vermont. (Virtually all HPHC Medicare beneficiaries 
reside in one of these states.) We merged this information onto the denominator file for 
Medicare beneficiaries and constructed three alternatives for the HPHC catchment area: 

(1) Four-state area : Beneficiaries living in Vermont, New Hampshire, Massachusetts, 
and Rhode Island. This catchment area produced a sample of 1 ,026, 1 83 beneficiaries meeting '" 
the basic fee-for-service sample criteria. 

(2) HPHC catchment area : Beneficiaries in a "3 digit zip code" in which HPHC had 
at least 10 Medicare members during the time period. The "3 digit zip code" is based on 
areas with the same first 3 digits of their zip codes, which tend to be clustered together 
geographically. For example, zip codes which begin with "021" are clustered in Boston's 
western suburbs. This definition results in a geographic area of southern Vermont, southern 
New Hampshire, eastern Rhode Island, and Massachusetts excluding a central region. This 
geographic limitation reduced the fee-for-service sample to 75 percent of that found above. 

(3) Refined HPHC catchment area : We constructed a map of zip codes in which 
HPHC Medicare beneficiaries reside (See Exhibit 2-1). Zip codes were classified into three - 
categories: those in which 1 to 5 HPHC Medicare beneficiaries reside, those with 6 to 29 
beneficiaries, and those with over 30 beneficiaries. The map indicates there are two major 
catchment areas: one surrounding Boston, and one in the far western portion of 
Massachusetts. The eastern catchment area corresponds fairly closely to the Boston CMSA. 
However, it does not extend as far west (Worcester area) as the CMSA or include New 
Bedford to the south. (These areas correspond to the location of HPHC providers.) We have 
constructed a comparison area that consists of all contiguous zip codes that comprise these 

Health Economics Research, Inc. Access in Managed Care Plans: 2-6 

crimson\final\chap2.wpd\dpb 



Chapter 2 



Methods 



Exhibit 2-1 

Zip Codes in which HPHC Medicare Beneficiaries Reside 



•vQ 



Vermont 





<§ New Hampshire 



If -^ 
Massachusetts 




Beneficiaries 

lto5 
6 to 29 
Over 30 



Health Economics Research, Inc. 

crimson\final\chap2.wpd\dpb 



Access in Managed Care Plans: 2-7 



Chapter 2 Methods 

catchment areas in which at least one HPHC Medicare beneficiary resides. This geographic 
limitation reduces the sample to 37 percent of the original 4 state sample. 

To determine the effect of varying the catchment area, we constructed one indicator, 
breast cancer screening rate, using each of the three definitions. We found that, compared 
to the four-state area (with a rate of 39.9 percent), the utilization rate was 0.2 percentage l 
points higher using the 3-digit zip code area and 0.9 percentage points higher using the 
refined catchment area. Although the results from the different catchment areas are quite 
similar, we feel the refined area best matches fee-for-service and HPHC beneficiaries, and 
we used this geographic definition to construct each of the indicators. 

2.3 Constructing Annual Indicators 

Several of our indicators are constructed as annual population-based utilization rates. 
Namely, the admission rate for ambulatory care sensitive conditions and all of our surgical 
(specialists) procedures are constructed as annual rates with the eligible sample as the 
denominator. One option for determining the sample would have been to include all - 
beneficiaries who met the eligibility criteria, using the first 12 months for which they were 
eligible in our time frame. However, for these indicators, we calculated a utilization rate for 
1994 basing the denominator on all individuals who were eligible for all of the calendar year. 
We then calculated the utilization rate for 1995, again basing the denominator on all 



Health Economics Research, Inc. Access in Managed Care Plans: 2-8 

crimson\final\chap2.wpd\dpb 



Cha P ter2 , Methods 

individuals eligible for the entire calendar year. We then averaged the two years. 5 (Very few 
HPHC beneficiaries were "lost" because they belonged to the plan for 12 consecutive 
months, but neither calendar year.) For this report, we present the averaged values in the 
text, and the annual rates in Appendix Table B-l . 

2.4 Adjusted vs. Unadjusted Rates 

The aged Medicare population enrolled in HPHC is substantially younger than the 
overall Medicare population. Nationally, 56 percent of aged beneficiaries are aged 65-74, 
compared to 70 percent of the HCD enrollees and 68 percent of the MGD enrollees. In 
addition, nationally 1 1 percent of aged beneficiaries are age 85 and older, compared to 4 
percent in the HCD and 7 percent in the MGD. 

In addition to the unadjusted rates, we present age-adjusted rates for HCD and MGD, 
standardized according to the proportion of patients in the fee-for-service sample in three 
groups: age 65-74, age 75-84, and age 85 and older. We also present the indicator for each 
of these three age groups, although the small sample sizes for many of the indicators result * 
in relatively large confidence intervals for the age-specific rates. Breakdowns by other 
demographic factors, such as race, were not constructed because of small sample sizes. 

Our performance indicators were developed to rely primarily on tracer conditions, 
with relatively well-defined populations in need of care. Use of tracer conditions allows us 



This approach simplifies indicator construction somewhat, since beneficiaries do not have to be tracked across 
multiple years (many claims-based files are constructed annually.) It also allows us to examine, for a subset of 
indicators, how the level of performance varies over a two-year period. 

Health Economics Research, Inc. Access in Managed Care Plans: 2-9 

crimson\final\chap2.wpd\dpb 



Cha P ter2 Methods 

to tie care to established clinical standards and greatly reduces the heterogeneity of the 
denominator population. Some of our indicators require no risk adjustment within the tracer 
group. For example, follow-up of some type is recommended for all women with abnormal 
mammograms. For some indicators (e.g., ambulatory care sensitive admission rates), it 
would be desirable to adjust for severity of the patient population. However, given the '" 
complexity of adequately risk-adjusting, especially given the different coding schemes used 
in the different data systems, we only perform the age-adjustment. 

2.5 Comparison with External Benchmarks 

For a number of our indicators, we are able to compare our results with external 
benchmarks. These benchmarks may take the form of goals that have been stated for access, 
such as the Healthy People 2000 objectives (DHHS, 1991), or previous studies which have 
constructed similar indicators, such as PPRC (1995) or the DHHS Report to Congress 
(1994). We do not compare our results with those from HEDIS, since previous HEDIS 
reports have been based on the under-65 population, while our study includes only the " 
elderly. 



Health Economics Research, Inc. Access in Managed Care Plans: 2-10 

crimson\final\chap2.wpd\dpb 




Indicator Specific 

Methods, 

Construction, and 

Results 



In this chapter we define each of the performance indicators and discuss the 
methodological issues faced in their construction. We also present results for each of the 
three sectors (fee-for-service, HCD, and MGD) overall and stratified by age. Table 3-1 
presents summary values for each indicator. (The 95 percent confidence interval is presented 
in parentheses below each indicator value.) For those wishing more detail on sample sizes 
and breakdowns by year, Appendix B presents more complete data on the rates for every 
indicator. 

3.1 Preventive Care 

3.1.1 Breast Cancer Screening Rate 

Definition 

Percentage of female beneficiaries receiving a mammogram during a 24-month 
period. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-1 

crimson\final\chap3 . wpd\dpb 



Chapter 3 



Results 



Table 3-1 
Age- Ad justed Summary of Performance Indicators 



Preventive Care 

Breast Cancer Screening Rate 
Percentage of female beneficiaries receiving a 
mammogram during a 24 month period 

Colon Cancer Screening Rate 
Percentage of beneficiaries with a fecal occult blood test 
sigmoidoscopy, or colonoscopy during a 24 month period 

Chronic Disease Care 

Rates of Secondary Preventive Services for Diabetes Mellitus 
Percentage of beneficiaries with a diabetes diagnosis 
with each of the following during a 12 month period: 

Retinal examination 



Two or more visits with a primary care provider or 
endocrinologist 

Population-Based Admission Rate for Ambulatory Care 
Sensitive Conditions 
Admission rates per 1,000 eligibles during a 12 month period 

Rate of Pre-Hospital Care for Ambulatory Care 

Sensitive Admissions 

Percentage of beneficiaries with an ACS admission with at 
least one visit during the 60 days prior to admission 

Rate of Post-Hospital Care for Ambulatory Care 

Sensitive Admissions 
Percentage of beneficiaries with an ACS admission with at 
least one visit during the 30 days following discharge 

Anti-hypertensive Follow-up Rate 
Percentage of beneficiaries with at least one follow-up 
visit within 8 months after receiving a prescription for an 
anti-hypertensive 



HPHC 



Fee-For 
Service 



Health Medical 

Centers Groups 
Division Division 



40.8% 77.0% 64.8% 

(40.6,41.0) (75.0,79.0) (61.6,68.0) 

35.8% 58.6% 52.7% 

(35.6,36.0) (57.5,60.7) (50.5,54.9) 



54.8% 67.5% 63.9% 

(54.3,55.3) (64.7,70.3) (59.6,68.2) 

61.2% 94.6% 90.7% 

(60.7,61.7) (92.5,96.7) (90.0,91.4) 

71.9 60.1 44.4 

(71.1,72.7) (55.7,64.5) (38.4,50.4) 

80.3% 85.8% 85.3% 

(79.7,80.9) (81.7,89.9) (79.1,91.5) 



78.4% 81.8% 84.6% 

(77.8,79.9) (77.3,86.3) (78.5,90.7) 



93.0% 
(92.1,93.9) 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-2 

crimson\final\chap3tabs\Tab3-1 \ss 



Chapter 3 



Results 



Table 3-1 (continued) 
Age-Adjusted Summary of Performance Indicators 



HPHC 



Fee-For 
Service 



Health Medical 

Centers Groups 
Division Division 



Anti-depressant Follow-up Rate 

Percentage of beneficiaries with at least one follow-up 
visit within 8 months after receiving a prescription for an 
anti-depressant 

Diagnosis-Specific Care 

Rate of Post-Hospital Follow-up for Myocardial Infarction 
Percentage of beneficiaries hospitalized for MI 
with at least one cardiology or primary care visit within 
60 days of discharge 

Rate of Post-Hospital Follow-up for Depression 
Percentage of beneficiaries hospitalized for 
depression with at least one primary care or mental 
health visit within 14 days of discharge 

Rate of Follow-up for Abnormal Mammogram 
Percentage of female beneficiaries with an abnormal 
mammogram who receive repeat mammogram, ultrasound, 
biopsy or surgery within 1 5 days 

Specialty Referral Care 

Population-based Rate of Lens Replacement 

Rate of lens replacements per thousand beneficiaries 
during a 12 month period 

Population-based Rate of Hip and Knee Replacement 
Rate of total hip and knee replacement per thousand 
beneficiaries during a 12 month period 

Population-based Rate of Coronary Revascularization 
Rate of coronary bypass and angioplasty per thousand 
beneficiaries during a 12 month period 

Rate of Breast Cancer Oncology Follow-up 
Percentage of female beneficiaries with at least one 
oncology or general surgery visit in the 6 months 
following an initial diagnosis of breast cancer 



93.2% 
(91.7,94.7) 



73.3% 90.7% 93.2% 

(71.7,74.9) (83.8,97.6) (83.8,100) 



65.8% 64.5% 80.3% 

(62.7,68.9) (37.3,92.2) (57.6,100) 



46.1% 
(34.2, 58.0) 



37.9 32.5 16.6 

(37.3,38.5) (29.1,35.9) (12.6,20.6) 

6.8 5.9 7.7 

(6.5,7.1) (4.3,7.5) (4.9,10.5) 

8.6 7.6 4.1 

(8.3,8.9) (5.7,9.5) (1.7,6.5) 



71.0% 
(63.7, 78.3) 



Health Economics Research, Inc. 

crimson\final\chap3 . wpd\dpb 



Access in Managed Care Plans: 3-3 



crimson\final\chap3tabs\Tab3- 1 \ss 



Chapter 3 



Results 



Table 3-1 (continued) 
Age-Adjusted Summary of Performance Indicators 



HPHC 



Fee-For 
Service 



Health Medical 

Centers Groups 
Division Division 



Primary Care 

Rate of New Enrollees with a Visit 
Percentage of new enrollees with at least one visit 
during the first two months of enrollment 

Rate of Beneficiaries with a Visit 
Percentage of beneficiaries with at least one visit with a 
primary care physician or specialist during a 12 month period 

Continuity of Care Index 

Proportion of visits per patient for primary care that are with 
the patient's primary care physician 



73.9% 48.7% 

(70.7,77.1) (45.1,52.3) 

88.4% 93.9% 90.9% 

(88.3,88.5) (91.7,96.1) (88.3,93.5) 



71.3% 
(66.7, 75.9) 



NOTE: Medicare fee-for-service did not cover routine colon cancer screening during the study period. Our rate 
may undercount the proportion of beneficiaries receiving the service if they paid out of pocket. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-4 

crimson\final\chap3tabs\Tab3-1\ss 



Cha P ter3 Results 



Data Specifications 

Denominator: Female aged beneficiaries enrolled continuously for the 24-month study 

period. 

Numerator: Those with a mammogram (CPT = 76091, 76092; AMRS = v 

R03 5— Mammogram [in use through August 1994]; R3 40— Mammogram Bilateral [in use 

starting August 1994] and R341 — Mammogram Unilateral [in use starting August 1994]). 

HPHC Indicator Construction 

Enrollment of women in the Medicare population in HPHC was initially analyzed 
from the MUPS demographic files. From the MUPS file, we identified female members in 
HPHC age 65 and over for some or all of the study period. We then identified those who 
were enrolled for 24 months continuously within one division during the period January 1994 
to December 1995. We then analyzed a 50 percent random sample from each division, 
leaving us with samples of 1,638 women in the HCD and 910 women in the MGD. 1 

HCD: From the AMRS we downloaded all encounters during the study period 
(January 1, 1994 to December 31, 1995) for the women in the denominator which included 
one of three test codes for mammography. 



' The nature of the encounter data makes processing of large samples for the HCD prohibitively time-consuming since 
it is an automated medical record and not a claims-based file. Although processing the entire MGD file is not substantially 
more expensive than processing a sample, a 50 percent random sample was drawn, analogous to the procedure used for the 
HCD. The resulting random samples are still much larger than the samples we found for many of our diagnosis-specific 
indicators. 

Health Economics Research, Inc. Access in Managed Care Plans: 3-5 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

MGD: From dummy claims we searched for CPT-4 codes for mammography during 
the two-year study period. Mammography was one of the first two indicators we constructed. 
Initial attempts to locate mammogram codes for the MGD identified only 2 percent of 
women with claims for a mammogram in a two-year period, including no claims in 1994. 
Investigating further, we discovered that claims for SeniorCare and Plan 65, the insurance '* 
products which enrolled Medicare recipients until 1995, were located not in the Ambulatory 
Claims Files, but rather in hospital claims files under an identifying code for SeniorCare. We 
included all claims from both the ambulatory and the institutional files in constructing the 
indicator. 

Fee-for-Service Indicator Construction 

From the denominator file, we identified women meeting the sample eligibility 
criteria (age 65 or older, continuously enrolled in fee-for-service) for the entire 24-month 
study period, yielding a sample of 211,026. We then searched physician/supplier and 
hospital outpatient claims files for claims with CPT-4 codes for mammography during the . 
two-year study period. Although the vast majority of mammography claims were located in 
the physician/supplier file, we found that 58 percent of women with a mammogram had 
claims in both the physician/supplier and hospital outpatient department files, and 4 percent 
had a claim in the outpatient department file but no physician/supplier claim. A claim from 
either file was taken as evidence that the woman had undergone a mammogram. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-6 

crimson\final\chap3 wpd\dpb 



Chapter 3 



Results 



Results 

From the three data sets, the percentages of women who had a mammogram 
performed during the 24-month period were: 



BREAST CANCER SCREENING RATE 

(Confidence Intervals 
in Parentheses) ,-, * „ 

Fee-for-Service 




HPHC 




Health Centers 
Division 


Medical Groups 
Division 


All eligibles (age 
adjusted) 


40.8% 

(40.6,41.0) 


77.0% 

(75.0, 79.0) 






64.8% 

(61.6,68.0) 


Age 65-74 


53.7 

(53.4, 54.0) 


86.6 

(84.4, 88.8) 






78.1 

(74.2,82.0) 


Age 75-84 


35.9 

(35.6, 36.2) 


76.6 

(73.1,80.1) 






61.2 

(56.0,66.4) 


Age 85 and older 


12.2 
(11.8, 12.6) 


46.7 

(37.1,56.0) 






31.1 
(21.0,41.2) 



The rate of mammography screening in the HCD is substantially higher than for the 
MGD, 77 percent compared to 65 percent. The uniform reminder system used by the HCD 
to prompt providers to offer annual screening mammography may help explain its 
exceptionally high rate and the difference between the two divisions. HCD and MGD are 
both performing above the goal for mammography set out in Healthy People 2000, the 
objectives set for the Nation's health into the next century. (That goal is for 60 percent of 
women age 50 and older to receive a mammogram within the previous one to two years). 
The rate in fee-for-service of 41 percent is similar to the mammography rate during a 24- 
month period of 38.6 percent nationally among the elderly found by PPRC (1995) and the 
39.3 percent found by HCFA (1998). 



Health Economics Research, Inc. 

crimson\final\chap3 . wpd\dpb 



Access in Managed Care Plans: 3-7 



Chapter 3 



Results 



We were also interested in examining variation in performance for our indicators 
within the two divisions of HPHC. Although the small samples and large number of 
practices make it impossible to look for meaningful differences among the groups of the 
MGD, we were able to construct mammography rates for each of the 14 centers of the HCD, 
as shown below. 



Center 


Sample 


Proportion Receiving 
Mammography 


A 


33 


78.8% 


B 


149 


75.8 


C 


140 


76.4 


D 


143 


67.8 


E 


19 


84.2 


F 


224 


83.6 


G 


221 


84.7 


H 


83 


81.4 


I 


42 


72.4 


J 


41 


78.9 


K 


189 


86.8 


L 


125 


74.4 


M 


98 


85.7 


Total 


1,638 


80.1 



The proportion of women receiving mammography in each center is quite high. 
Seven of the 14 centers had rates exceeding 80 percent. The lowest rate is 67.8% for Center 
D, which is still well above the rate in fee-for-service or in the MGD (as well as the Healthy 



Health Economics Research, Inc. 

crimson\final\chap3 . wpd\dpb 



Access in Managed Care Plans: 3-8 



Chapter 3 Results 

People 2000 objective). Because of the small sample sizes for many centers, we did not 
construct confidence intervals around each of the rates. However, a chi-square test for 
differences in proportions across all centers was significant at the one percent level. 
3.1.2 Colon Cancer Screening Rate 
Definition 

Percentage of beneficiaries receiving a fecal occult blood test, sigmoidoscopy, or 
colonoscopy during a 24-month period. 

Data Specifications 

Denominator : Aged beneficiaries enrolled continuously for the 24-month study period. 
Numerator : Those with a fecal occult blood test (CPT= 82270; AMRS = TY150, Q700), 
stool occult blood test, colonoscopy or sigmoidoscopy (CPT = 45300, 45305, 45308, 45309, 
45315, 45320, 45330, 45331, 45332, 45338, 45339, 45378, 45380, 45383, 45384, 45385; 
AMRS = W0 13 - sigmoidoscopy; T073 - colonoscopy, diagnostic; T074 - colonoscopy, for 
biopsy; T075 - colonoscopy, for stricture dilation; T076 - colonoscopy, for polypetctomy; 
T077 - colonoscopy - for control of hemorrhage; T078 - flexible sigmoidoscopy, for biopsy; 
T079 - flexible sigmoidoscopy, for polypectomy; T080 - flexible sigmoidoscopy, for ablation 
of tumor; T081 - rigid sigmoidoscopy, diagnostic; T082 - proctosigmoidoscopy, for biopsy; 
TO 8 3 - rigid sigmoidoscopy, anoscopy; K404 - negative sigmoidoscopy exam; T386 - 
sigmoidoscopy, diagnostic; T387 - sigmoidoscopy, for removal of colonic polyp; T549 - 
flexible sigmoidoscopy, diagnostic; T569 - proctosigmoidoscopy, direct; Y144 - 



Health Economics Research, Inc. Access in Managed Care Plans: 3-9 

crimson\final\chap3.wpd\dpb 



Cha P ter3 Results 

sigmoidoscopy with biopsy, Y303 - sigmoidoscopy; Y486 - colonoscopy, Y489 - 
colonoscopy and polypectomy). 

HPHC Indicator Construction 

From the MUPS enrollment file, we identified members in HPHC age 65 and older * 
who were enrolled for 24 months continuously within one division during the period January 
1994 to December 1995. We then took a random sample from each division, leaving us with 
2,089 beneficiaries in the HCD and 2,045 beneficiaries in the MGD. 

HCD: From the AMRS we downloaded all encounters during the study period 
(January 1, 1994 to December 31, 1995) for beneficiaries in the denominator which included 
one of the test codes for colorectal cancer screening. 

MGD: From dummy claims we searched for CPT-4 codes for colorectal cancer 
screening during the two-year study period. 

Fee-for-Service Indicator Construction 

From the denominator file, we identified individuals meeting the sample eligibility 
criteria for the entire study period, resulting in a sample of 339,627. We then searched 
physician/supplier claims files for claims with CPT-4 codes for fecal occult blood test, 
colonoscopy, or sigmoidoscopy during the two year study period. Individuals receiving any 
of the three procedures were considered to have received a colon cancer screening test. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-10 

crimson\final\chap3 . wpd\dpb 



Chapter 3 . Results 

Results 

The rates of colon cancer screening, by fecal occult blood test, sigmoidoscopy or 
colonoscopy among the three groups were: 



COLON CANCER 
SCREENING RATE 


Fee-for-Service 


HPHC 


Health Centers 
Division 


Medical Groups 
Division 


All eligibles (age 
adjusted) 


35.8% 

(35.6, 36.0) 


58.6% 

(57.5, 60.7) 


52.7% 

(50.5, 54.9) 


Age 65-74 


38.1 

(37.9, 38.3) 


63.1 

(60.4, 65.8) 


57.5 
(54.7, 60.3) 


Age 75-84 


35.9 
(35.6, 36.2) 


60.4 

(56.7,64.1) 


51.9 

(48.2, 55.6) 


Age 85 and older 


25.4 
(25.0, 25.8) 


33.3 

(22.6, 44.0) 


34.6 
(26.0, 43.2) 



The rate of colon cancer screening is substantially higher in the HCD and the MGD 
than the fee-for-service sector. Although the goals set out in Health People 2000 are not 
identical to our indicator, both divisions of the HMO appear to be performing above this 
objective. (That goal was for 50 percent of the population age 50 and older to have received 
fecal occult blood testing within the preceding one to two years.) In contrast, fee-for-service, 
with a screening rate of 35.8 percent, is well below that objective. However, colon cancer 
screening was not routinely covered in fee-for-service until 1988. Thus, some proportion of 
beneficiaries may have received the service but had no claims appear in our data. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-11 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

3.2 Chronic Disease Care 

3.2.1 Retinal Examination Rate for Diabetes 

Definition 

Percentage of beneficiaries with a diabetes diagnosis receiving a retinal screening 
examination during a 1 2-month period. 

Data Specifications 

Denominator : All aged beneficiaries continuously enrolled for at least 12 months in the 
study period following the appearance of a diagnosis of diabetes mellitus (ICD-9 = 250, 
AMRS = B120). 2 

Numerator : Those receiving a retinal exam (CPT = 92002-92014, 92225, 9226) or having 
a procedure or diagnosis code that suggests a dilated retinal exam was performed (AMRS 
= T589-retina: prophylaxis, cryotherapy; T590-retina: prophylaxis, photocoagulation; T591- 
retina: destruction of retinal lesion; T592-retina: destruction of retinopathy; T643-ocular 
photography: fluorescein angiography; T630-ophthalmoscopy with fundus photography; 
D115-retinal scars; D146-retinal arteriolar sclerosis; D149-retinal defect; D160-branch 
retinal vein occlusion; D164-retinal hole or tear; D165-central retinal artery occlusion; D208- 
central retinal vein occlusion; D209-epiretinal membrane; D228-branch retinal artery 
occlusion; D23 1 -retinal vasculitis; D234-commotio retinae; D544-retinal vascular occlusion; 
D550-retinopathy; D551 -arteriosclerotic retinopathy; D552-diabetic retinopathy; D553- 



2 For example, a beneficiary with a diagnosis appearing in March 1994 would be followed for 12 months (including part 
of 1995). 

Health Economics Research, Inc. Access in Managed Care Plans: 3-12 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

hypertensive retinopathy; D554-detached retina; D555-retinal degenerative disease; D562- 
diabetic retinopathy, background; D563-diabetic retinopathy, proliferative; D163-central 
serous retinopathy; D5 1 0-chorioretinitis; D161 -chorioretinitis from toxoplasmosis; D212- 
choroidal nevus; D162-drusen; D003-floaters; D558-lattice degeneration; D530-macular 
degeneration; D 1 82-macular edema; D2 1 1 -macular hole; D 1 66-ocular histoplasmosis; D222- k 
posterior vitreous detachment; D883-retinal detachment; D287-retinitis; D204-retinitis 
pigmentosa; D906-retinoschisis; D129-rubeosis iridis; D172-vitreous hemorrhage; D242- 
myelinated nerve fibers; D340-optic atrophy; D170-optic disc edema; D218-optic 
nervedrusen; D171-pseudopapilledema; D026-eye examination, normal, with modifier SET 
175); D184-choroidal atrophy; D109-drusen; D600-uveitis; D140-optic disc drusen; D543- 
neuritis; D0290-papilledema; D155-pseudophakia; D177-pseudoexfoliative syndrome; 
D181-opaque posterior capsule; D4 1 0-cataract; D450-iritis; D704-aphakia). 
HPHC Indicator Construction 

HCD: For the HCD, 1 ,325 eligibles (12 percent) were found to have a diagnosis code 
for diabetes used in at least one encounter within the first year of the study period. This - 
number was further refined to those who were enrolled for twelve months continuously 
following that diagnosis, yielding a sample of 1,239. These were further limited to those 
who had the diagnosis used during a face to face visit, and for whom the code was designated 
as "major," "minor" or "dictation" (i.e., not "rule out" or "presumptive") resulting in a 
sample of 1,092. These last requirements brought the definition of the denominator in the 
HCD into parallel with the definition for the MGD and fee-for-service data, where claims 
would only be generated for a face to face visit. 

Health Economics Research, Inc. Access in Managed Care Plans: 3-13 

crimson\final\chap3 . wpd\dpb 



Cha P ter3 Results 

We had originally proposed using the AMRS codes for retinal examination (AMRS 
= Dl 15, D131) to identify patients receiving the exam in the HCD. 3 However, physicians 
in this division have little incentive to code that the exam was completed (since 
reimbursement is not dependent on coding), and we found that they often coded a diagnosis 
which would normally require a retinal exam without coding the exam itself. Thus, we * 
developed the list of codes that imply a retinal exam has been performed. For comparison, 
we also calculated the proportion of diabetics with a visit to an optometrist or an 
ophthalmologist. 

In the HCD 67 percent of diabetics had a face to face visit to an eye specialist who 
charted a diagnostic code (e.g. diabetic retinopathy, vascular occlusion) that suggests a 
dilated retinal examination was performed, while roughly 77 percent of diabetics had a face 
to face encounter with an optometrist or an ophthalmologist. Significant effort went into 
defining the reason for the difference between these measures. Provider specialty is 
identified by the department in which the patient is seen and the characterization of this 
variable in the data set is excellent. Optometrists and ophthalmologists both provide primary - 
eye care for adults in the HCD. 4 When an ophthalmologist or optometrist sees someone who 
is diabetic for whatever reason, one expects that they would include a retinal examination. 
However, there is no way to know if that is true from the AMRS data, because the eye 
doctors maintain a separate chart on paper, which includes their drawing of the retinal 



3 D131 — Routine eye exam was not included in the codes used to identify patients. 

"In the fee-for-service world, optometrists mostly do refractions and fit eye glasses. In rural areas, they maybe more likely 
to do retinal exams. 

Health Economics Research, Inc. Access in Managed Care Plans: 3-14 

crimson\final\chap3 . wpd\dpb 



Cha P ter3 , Results 

findings. Given this separate chart and the lack of any incentive to code "screening dilated 
retinal exam" every time one is done, the clinician is more likely to code what he or she 
sees, i.e. the abnormality, in the AMRS data. However, it is important to note that we may 
be undercounting patients (if some had no abnormality coded), or over counting patients (if 
some had a diagnosis coded, but no retinal exam performed). Since 77 percent of patients * 
saw an optometrist/ophthalmologist, this is the upper limit on the number that could have 
received the exam. 

The list of diagnostic codes that we used was created by an internist's review of all 
possible D-codes (diagnostic codes) in AMRS and then review of additional D-codes used 
by ophthalmologists and optometrists. The ophthalmologist who consulted on these codes 
thought that, because ophthalmologic exams within the HCD are charted freehand on paper, 
and thus not completely automated, a more precise estimate for this indicator would require 
chart review. 

MGD: From the MGD, 587 eligibles (8 percent) have been identified as diabetic 
from at least one ICD-9 code from an ambulatory or inpatient visit during 1994. Of these, 
495 remained in the plan for twelve months continuously following the documentation of 
diabetes. 5 

In the MGD, we searched claims for a CPT-4 code indicating that a retinal exam had 
been performed in the 12 months following the first appearance of a diabetes diagnosis. We 



5 For comparison, we also attempted to identify diabetics using pharmacy data. For the period 1993-1995, the pharmacy 
approach identified 539 diabetics, while the diagnosis approach found 630 (virtually all of whom were found using the 
pharmacy approach). In the end, we did not use the pharmacy information because we did not have comparable information 
available for the other HPHC division or fee-for-service beneficiaries. 

Health Economics Research, Inc. Access in Managed Care Plans: 3-15 

crimson\final\chap3.wpd\dpb 



Cha P ter3 , _ Results 

found that 61 percent of diabetics had a claim coded with a specific CPT-4 code for 
screening retinal exam; 45 percent had a claim for a visit to an optometrist or an 
ophthalmologist. The lower number of visits to "ophthalmologists" than retinal exams in the 
MGD suggests the possibility of imprecision in the Provider Specialty field for these claims. 
(Almost all claims for a retinal exam not coded as ophthalmologist/optometrist were coded 
as specialty "unknown".) While this indicator is based on the presence of the CPT-4 code, 
the imprecision of Provider Specialty in the MGD file may be important for other indicators 
that key off this variable. 

Fee-for-Service Indicator Construction 

We determined the denominator for this indicator by searching the physician/supplier 
claims files for appearance of the appropriate ICD-9 diagnosis codes for diabetes on a 
physician claim. (We excluded laboratory claims because of the high rate of diagnostic 
coding for rule-out of diabetes, and to be consistent with indicator construction in HPHC.) 
This yielded a sample of 34,260 beneficiaries (9 percent of eligibles) who remained in the 
sample for 12 months following the first documentation of diabetes. 

For these individuals, we searched claims for a CPT-4 code indicating that a retinal 
exam had been performed in the 12 months following the first appearance of a diabetes 
diagnosis. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-16 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



Results 

The rate of diabetic retinal examination within a twelve-month period following the 
initial diagnosis of diabetes in our data, for each of the three groups was: 



RETINAL EXAM RATE 






HPHC 




Health Centers 


Medical Groups 




Fee-for-Service 


Division 






Division 


All eligibles (age 


54.8% 


67.5% 






63.9% 


adjusted) 


(54.3, 55.3) 


(64.7, 70.3) 






(59.6, 68.2) 


Age 65-74 


53.1 


64.8 






61.6 




(52.3, 53.9) 


(61.3,68.3) 






(56.0, 67.2) 


Age 75-84 


57.3 


68.9 






67.1 




(56.5,58.1) 


(63.7,74.1) 






(60.2, 75.2) 


Age 85 and older 


52.2 


76.0 






62.1 




(50.4, 54.0) 


(57.3, 94.7) 






(42.7,81.5) 



These rates suggest that in both HMO settings the completion of annual retinal 
screening for diabetics falls considerably short of the goal of 100 percent, but their 
achievement exceeds findings in fee-for-service practice. Our fee-for-service sample had a 
rate somewhat lower than that found in the two HMO divisions. However, in previous 
studies three states demonstrating the use of the Delmarva indicators had an overall annual 
rate of eye exams for diabetics of 45.9 percent (JAMA, 1995) and PPRC (1995) found a rate 
of 38.2 percent. Thus, our fee-for-service rate is noticeably higher than that found in other 
studies. Our higher fee-for-service utilization rate may result from use of a sample that is 
primarily urban, based in the Boston metropolitan area. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-17 



Cha P ter3 , . Results 

We required only one diabetes diagnosis for a beneficiary to be included in the 
sample for this indicator, since we did not want to leave out diabetics who had little contact 
with the health care system. Other studies, including HEDIS, require two diagnoses, in 
hopes of eliminating patients who had been coded with a rule-out diagnosis of diabetes. To 
test the sensitivity of the measure to definition of the sample, we also constructed the rate of '** 
the retinal exam for patients with two or more physician diagnoses in a twelve month period. 
The effect on the rate of retinal exam is minor, increasing it slightly in HCD and MGD, and 
decreasing it in fee-for-service. The effect on the sample size is more noteworthy. In fee- 
for-service the sample decreases by 38 percent, in the MGD by 20 percent, and in the HCD 
by 15 percent. The large proportion of the sample lost in fee-for-service is consistent with 
the hypothesis that many diabetes diagnoses are "rule outs". However, the 1 5 percent sample 
reduction in the HCD sample is more puzzling. Given the HCD coding system, a diagnosis 
of diabetes should only be found for patients with confirmed disease. That 1 5 percent of 
diabetics have only one diagnosis may reflect rule outs that were not coded as such, 
miscodes, failure to code the diagnosis for every visit, or that the patient had only one visit - 
during the year. 

3.2.2 Visit Rate for Diabetes 

Definition 

Percentage of beneficiaries with a diabetes diagnosis with two or more visits with a 
primary care provider or endocrinologist during a 12-month period. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-18 

crimson\final\chap3 wpd\dpb 



Chapter 3 __ Results 

Data Specifications 

Denominator : Aged beneficiaries continuously enrolled for at least 12 months in the study 
period following the appearance of a diagnosis of diabetes mellitus (ICD-9 = 250). 
Numerator : Those with two or more visits with a primary care provider or endocrinologist 
during the 12-month study period. 

HPHC Indicator Construction 

The denominator for this indicator was identical to that for diabetic retinal exam 
(Section 3.3), namely, those with a diagnosis of diabetes who remained in the plan for 12 
consecutive months during the study period. 

HCD: The characterization of specialty in the HCD files is excellent. We searched 
the AMRS for claims indicating that the individual had two or more visits (on different 
dates) with an internist or endocrinologist during the twelve months following the original 
diagnosis of diabetes. (HCD does not employ general/family practitioners as primary care 
physicians.) The AMRS data does not include inpatient and emergency room utilization, so 
it was not necessary to explicitly exclude these from the data. 

MGD: Specialty characterization is less clean in the MGD. Specifically, as part of 
the SeniorCare/Plan 65 product, visits to medical groups frequently coded provider type as 
"Institution-Interdivisional Care" which does not specify level of professional (i.e., M.D., 
N.P.) or the specialty of the provider seen. We used this provider type as a proxy for internal 
medicine visits, but it doubtless is less specific than HCD. Using this specification we 



Health Economics Research, Inc. Access in Managed Care Plans: 3-19 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

searched the data for two or more visits in an outpatient setting following the first appearance 
of the diabetes diagnosis. 

Fee-for-Service Indicator Construction 

The denominator for this indicator was identical to that for diabetic retinal exam k 
(Section 3.3), namely, those with a diagnosis of diabetes who remained in fee-for-service for 
12 consecutive months during the study period. 

We originally searched the physician/supplier claims file for claims with a specialty 
of general/family practice, internal medicine, endocrinology or geriatrics, and a place of 
service indicating office or outpatient clinic treatment. However, the results of this process 
yielded a surprisingly low number of patients with two or more visits during the year 
interval. Further examination of the data revealed that roughly 16 percent of diabetic claims 
were coded with a specialty of "multispecialty clinic or group practice" (compared with 1 2 
percent coded with a specialty of internal medicine). Among claims with the multispecialty 
group or clinic code, almost half contained evaluation and management CPT-4 procedure . 
codes. Thus, although we cannot tell the specially of the provider seen by the beneficiary, 
it appears likely that many of these visits were for primary care. As a result, the specialties 
included in constructing the indicator were expanded to include the multispecialty 
clinic/group code. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-20 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



Results 

The proportion of diabetics having at least two visits with a primary care provider or 
an endocrinologist within a year following the initial appearance of a diabetes diagnosis in 
our data was as follows: 



VISIT RATE FOR 
DIABETES 


Fee-for-Service 


HPHC 


Health Centers 
Division 


Medical Groups 
Division 


All eligibles (age 
adjusted) 


61.2% 

(60.7,61.7) 


94.6% 

(92.5, 96.7) 


90.7% 

(90.0,91.4) 


Age 65-74 


59.4 

(58.7,60.1) 


95.4 

(92.9, 97.9) 


91.5 

(89.4, 93.6) 


Age 75-84 


61.9 

(61.1,62.7) 


92.5 

(88.1,96.9) 


90.7 

(87.4, 94.0) 


Age 85 and older 


67.0 

(65.3, 68.7) 


100.0 

(93.7, 100.0) 


88.0 

(75.3, 100.0) 



These rates suggest that in both HMO settings the proportion of diabetics with at least 
two visits during a 12- month period is quite high. It is interesting to note that the visit rate 
is highest in the HCD, in which specialty can be accurately identified, and for which we are 
virtually certain that visits are with either a primary care physician or endocrinologist. Both 
fee-for-service and MGD data suffer from the coding of "groups" which contain unidentified 
or multiple specialties. Because of the prevalence of visits with this specialty code, and the 
appearance that much of what took place during these visits was primary care, we included 
them in our measures. To the extent that these "group" visits are with physicians other than 



Health Economics Research, Inc. 

crimson\final\chap3 . wpd\dpb 



Access in Managed Care Plans: 3-21 



Cha P ter3 ___ Results 

primary care physicians or endocrinologists, we would expect these results to be biased 
upwards compared to the HCD. 

3.2.3 Admission Rate for Ambulatory Care Sensitive Conditions 

Definition 

Admission rates during a 1 2 month period for selected diagnoses for which admission 
may be potentially preventable through the use of primary care. 

Data Specifications 

Denominator : Aged beneficiaries continuously enrolled for a 12-month period. 

Numerator : All admissions with a principal diagnosis of an ambulatory care sensitive (ACS) 

condition, defined as follows: 

Tuberculosis: ICD-9 = 011 

Chronic obstructive pulmonary disease: ICD-9 = 491, 492, 494, 496 

Pneumonia: ICD-9 = 481, 482, 483, 485, 486 

Asthma: ICD-9= 493 

Congestive heart failure: ICD-9 = 428, 402.01, 402.11, 402.91, 518.4 

Hypertension: ICD-9 = 401.0, 401.9, 402.00, 402.10, 402.90 

Angina: ICD-9 = 411.1, 411.8, 413 (and no procedure) 

Cellulitis: ICD-9 = 681, 682, 683, 686 

Kidney-urinary infections: ICD-9 = 590, 599.0 

Severe ENT infections: ICD-9 = 382, 462, 463, 465 

Other tuberculosis: ICD-9 = 012, 013, 014, 015, 016, 017, 018 

Diabetes with ketoacidosis or coma: ICD-9 = 250.1, 250.2, 250.3 

Diabetes with other complications: ICD-9 = 250.9, 250.7 

Diabetes with no complications: ICD-9 = 250.0 

Hypoglycemia: ICD-9 = 250.8 

Gastroenteritis: ICD-9 = 558.9 

Dehydration: ICD-9 = 276.5 

Nutritional Deficiencies: ICD-9 = 260, 261, 262, 268.0, 268.1 



Health Economics Research, Inc. Access in Managed Care Plans: 3-22 

crimson\final\chap3 .wpd\dpb 



Chapter 3 Results 



Grand mal status/epileptic convulsions: ICD-9 = 345 
Other convulsions: ICD-9 = 780.3 

The ambulatory care sensitive conditions are discussed in detail in Billings (1993). 



HPHC Indicator Construction 

From the MUPS enrollment file, we identified members in HPHC age 65 and older 
who were enrolled within one division for all of calendar year 1994 and/or all of calendar 
year 1995 (see Section 2.3 for a discussion of this sampling strategy). This yielded samples 
of 8,764 and 9,075 beneficiaries for 1994 and 1995, respectively, for the HCD and 4,196 and 
4,319 beneficiaries for 1994 and 1995, respectively, for the MGD. 

HCD: The institutional file was searched for admissions with an ACS principal 
diagnosis. Admission and discharge dates were analyzed to ensure that patients being 
transferred from one hospital to another were not being double counted. (If the discharge 
date from one facility was identical to the admission date at another facility, this was 
considered a transfer, and was counted as only one admission. However, the same 
beneficiary could have multiple admissions, as long as they did not meet the transfer 
criterion.) 

MGD: The institutional file was searched for admissions with an ACS principal 
diagnosis. Transfers were handled as in the HCD. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-23 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



Fee-for-Service Indicator Construction 

The denominator file was used to identify beneficiaries who met the sample criteria 
to be counted in the denominator for this indicator — enrollment for all of calendar year 1994 
and/or 1995. This yielded samples of 363,934 for 1994 and 329,116 for 1995. MedPAR 
inpatient admission files were then searched for hospitalizations with an ACS principal l 
diagnosis code. Admission and discharge dates were checked to ensure that transferred 
patients were not double-counted. 



Results 

We present only the admission rate for all ACS conditions in the aggregate, rather 
than the rate for each individual condition, because of small sample sizes within the HPHC 
data. Rates were calculated for each of the two years and then averaged. Rates are presented 
as admissions per 1 ,000 beneficiaries. 



ACS ADMISSION RATE 


Fee-for-Service 


HPHC 


Health Centers 
Division 


Medical Groups 
Division 


All eligibles (age 
adjusted) 


71.9 

(71.1,72.7) 


60.1 

(55.7, 64.5) 


44.4 

(38.4, 50.4) 


Age 65-74 


49.8 

(48.8, 50.8) 


30.0 

(25.5, 34.5) 


30.7 

(33.9, 39.5) 


Age 75-84 


85.2 
(84.7, 86.7) 


63.9 

(54.6, 73.2) 


50.6 

(38.8, 62.4) 


Age 85 and older 


126.1 

(122.7, 129.5) 


182.9 

(141.7,224.1) 


84.7 
(50.5, 118.9) 



Health Economics Research, Inc. 

crimson\fmal\chap3.wpd\dpb 



Access in Managed Care Plans: 3-24 



Chapter 3 Results 

The admission rates for fee-for-service beneficiaries are somewhat higher than those 
for the two divisions of HPHC. In previous studies, Mitchell (1994) found an ACS 
admission rate among the elderly of 71.8 per thousand beneficiaries in health professional 
shortage areas and 60.8 per thousand beneficiaries in nonshortage areas. Rosenbach and 
Khandker (1994) found a rate of 41 admissions per thousand beneficiaries. The reason for »• 
our somewhat higher rate is not obvious. However, it is consistent with previous work by 
Wennberg (1996) which found that the Boston area (where most of our beneficiaries reside) 
has very high rates of hospitalization for conditions such as pneumonia, COPD, and 
congestive heart failure, for which severity varies substantially across patients. He 
hypothesizes that this high admission rate is related to the high number of hospital beds per 
capita in the Boston area. 

3.2.4 Rate of Pre-hospital Care for Ambulatory Care Sensitive 
Admissions 

Definition 

Of patients with an ambulatory care sensitive admission, the percentage with at least 
one physician visit during the previous 60 days. 

Data Specifications 

Denominator : Those with an ACS admission during the study period, for whom we had data 

for 60 days prior to admission. (The admission was March 2, 1994 or later.) 



Health Economics Research, Inc. Access in Managed Care Plans: 3-25 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

Numerator: Those with at least one (non ER, non-inpatient) physician visit in the 60 days 
prior to admission. 

Some individuals have multiple ACS admissions (roughly 5 percent with an ACS 
admission have another during the year) so that the pre-admission period for one 
hospitalization may overlap with the post-admission period for another hospitalization. For l 
each beneficiary, we used the first ACS admission in each calendar year (allowing for a 60 
day pre-admission window in our data) in constructing this indicator. 

HPHC Indicator Construction 

HCD: The institutional file was searched for the first ACS admission for each 
beneficiary during each calendar year, allowing for 60 days of pre-hospitalization data. This 
yielded a sample of 311 admissions for 1994 and 310 admissions for 1995. For each 
admission, we then identified all physician claims for the 60 day period prior to admission. 
These were searched for claims indicating the beneficiary had a physician office visit during 
that period. 

MGD: We followed the same process as in the HCD to identify ACS admissions, 
yielding a sample of 128 for 1994 and 153 for 1995. Claims were then searched for 
physician office or outpatient visits during the 60 days prior to admission. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-26 

crimson\finaI\chap3.wpd\dpb 



Chapter 3 



Results 



Fee-for-Service Indicator Construction 

Our file with ACS admissions was searched for the first admission for each 
beneficiary during each calendar year, allowing for 60 days of pre-hospitalization data. This 
yielded a sample of 16,778 admissions for 1994 and 22,577 admissions for 1995. For each 
admission, we then identified all physician claims for the 60 day period prior to admission. 
These were searched for claims indicating the beneficiary had a physician visit during that 
period occurring in an office or outpatient setting. 



Results 

The proportion of patients with an ambulatory visit prior to their first ACS 
hospitalization was calculated for each year and then averaged. The results were as follows: 



ACS PREHOSPITAL 






HPHC 




CARE RATE 


Fee-For- 










Health Centers 


Medical Groups 




Service 


Division 






Division 


All eligibles (age 


80.3% 


85.8% 






85.3% 


adjusted) 


(79.7, 80.9) 


(81.7,89.9) 






(79.1,91.5) 


Age 65-74 


79.1 


82.6 






85.0 




(78.1,80.1) 


(76.2. 89.0) 






(76.4, 93.6) 


Age 75-84 


81.6 


90.9 






86.3 




(80.8, 82.4) 


(86.0, 95.8) 






(75.6, 87.0) 


Age 85 and older 


79.2 


82.8 






83.0 




(78.0, 80.4) 


(70.9, 94.7) 






(60.8, 100.0) 



Health Economics Research, Inc. 

crimson\final\chap3 wpd\dpb 



Access in Managed Care Plans: 3-27 



Chapter 3 



Results 



Our sample sizes for both divisions of HPHC are quite small for this indicator. 
However, it appears for all three sectors that the proportion of patients receiving care prior 
to the ACS admission is quite high. 

Outpatient care prior to an ACS admission is not an indicator that has been used in 
other studies. The indicator, as we have currently constructed it, simply measures whether * 
the person had any physician visits in the 60 days prior to admission. To test the sensitivity 
of the indicator to this specification, we also constructed four alternatives: proportion of 
patients with a visit in the 30 days or 7 days prior to admission, and proportion with a visit 
in the 30 days or 7 days prior to admission excluding the day prior to admission. These 
results are presented below: 



ACS PREHOSPITAL CARE 

ALTERNATIVE 

SPECIFICATIONS 


Fee-For- 
Service 




HPHC 




Health Centers 
Division 


Medical Groups 
Division 


60 Days prior 




80.3% 


85.8% 






85.3% 


30 Days prior 




68.0 


76.1 






61.0 


7 Days prior 




36.1 


54.4 






53.7 


30 Days prior (exclude day 
before admission) 


64.5 


65.2 






51.3 


7 Days prior (exclude 
before admission) 


day 


28.0 


29.8 






24.7 



As would be expected, the proportion of patients with a visit decreases substantially 
when the window is shortened from 60 to 30 or 7 days prior to admission. More interesting 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-28 



Chapter 3 Results 

is the effect of excluding visits the day before admission. When comparing visit rates the 7 
days before admission, fee-for-service (with a rate of 36 %) lags substantially behind both 
divisions of HPHC (with rates of 54%). However, when visits the day prior to admission are 
excluded, the fee-for-service visit rate at 28 percent is quite comparable to the 30 percent for 
the HCD and 25 percent for the MGD. (Since ACS admissions are not elective, we would »■ 
not expect visits the day before admission to be for planned pre-testing. Thus, the difference 
in rates excluding the day before admission vs. including this day do not reflect philosophical 
differences in pre-stay testing). 

3.2.5 Rate of Post-hospital Care for Ambulatory Care Sensitive 
Admissions 

Definition 

Of patients with an ambulatory care sensitive admission, the percentage with at least 
one physician visit during the 30 days following discharge. 

Data Specifications 

Denominator: Those with an ACS admission during the study period, who did not have a 

subsequent admission during the 30 days following the ACS discharge. 

Numerator: Those with at least one (non ER, non-inpatient) physician visit in the 30 days 

following discharge. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-29 

crimson\finaI\chap3.wpd\dpb 



Chapter 3 Results 

HPHC Indicator Construction 

HCD: The institutional file was searched for ACS admissions for each beneficiary 
during each calendar year, allowing for 30 days of post-discharge data. For each 
hospitalization we searched for another inpatient admission during the 30 days after 
discharge. This yielded a sample of 297 admissions for 1 994 and 276 admissions for 1 995 . '" 
We then identified all physician claims for the 30 days after discharge. These were searched 
for claims indicating the beneficiary had a physician office visit during that period. 

MGD: We followed the same process as in the HCD to identify ACS admissions and 
rehospitalizations, yielding a sample of 148 for 1994 and 140 for 1995. Claims were then 
searched for physician or outpatient visits during the 30 days following discharge. 

Fee-for-Service Indicator Construction 

Our file was searched for ACS admissions for each beneficiary during each calendar 
year, and patients with a rehospitalization within 30 days were removed from the sample. 
This yielded 13,895 admissions for 1994 and 18,249 admissions for 1995. For each - 
admission, we then identified all physician claims for the 30 day period after discharge. 
These were searched for claims indicating the beneficiary had a physician visit during that 
period occurring in an office or outpatient setting. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-30 

crimson\finaI\chap3 wpd\dpb 



Chapter 3 



Results 



Results 

The proportion of patients with an ambulatory visit after their ACS hospitalization 
was calculated for each year and then averaged. The results were as follows: 



ACS POST-HOSPITAL CARE 
RATE 


Fee-For-Service 




HPHC 


Health 
Centers 
Division 


Medical 

Groups 

Division 


All eligibles (age adjusted) 


78.4% 

(77.8, 79.9) 


81.8% 

(77.3, 86.3) 


84.6% 

(78.5, 90.7) 


Age 65-74 


80.2 

(79.1,81.3) 


87.8 

(81.4,94.2) 


89.6 

(81.7,97.5) 


Age 75-84 


79.1 

(78.2, 80.0) 


83.3 
(76.4, 90.2) 


83.3 

(72.1,94.5) 


Age 85 and older 


73.8 

(72.3, 75.3) 


68.4 

(54.1,82.7) 


79.3 

(62.0, 96.6) 



The rate of follow-up care was quite similar in all three sectors, ranging from 78.4 percent 
of fee-for-service patients to 84.6 percent of patients in the MGD. For all 3 sectors we 
dropped from the sample beneficiaries with a re-admission within 30 days of discharge. This 
led to a reduction of roughly 8 percent of the fee-for-service on a HCD samples. The MGD, 
with its smaller sample to start with, experienced a slightly greater attrition rate due to 
readmissions. This difference was not, however, statistically meaningful. 



Health Economics Research, Inc. 

crimson\final\chap3 wpd\dpb 



Access in Managed Care Plans: 3-31 



Chapter 3 Results 

3.2.6 Anti-Hypertensive Follow-Up Rate 

Definition 

Percentage of beneficiaries with a prescription for an ACE inhibitor or Loop diuretic 
with at least one outpatient visit to a primary care provider or cardiologist during the eight 
months after the prescription was written. 

Data Specifications 

Denominator: Aged beneficiaries enrolled in the HCD for at least eight months following the 

date a prescription was written for an ACE inhibitor or loop diuretic. 

Numerator: Those with at least one visit to internal medicine or cardiology during the eight 

months after receiving the prescription. 

HPHC Indicator Construction 

HCD: The AMRS was searched for patients with an appropriate prescription code 
during the period from January 1, 1993 to April 30, 1995. We then subset to patients for 
whom we had data for at least eight months following the date the prescription was written. 
(If a patient had more than one prescription, we used the first one in our sampling period.) 
This yielded a sample of 3,078 eligibles with an anti -hypertensive prescription. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-32 

crimson\final\chap3.wpd\dpb 



Cha P ter3 Results 

Results 

Below we present the rates of follow-up within eight months after receiving a 
prescription for an anti-hypertensive. We also present the rate of follow-up after six months 
to determine the effect of varying the length of the episode. 



ANTI-HYPERTENSIVE 






FOLLOW-UP RATE 


Health Centers Division 




6 Months 


8 Months 


All eligibles 


90.4% 


93.0% 


* 


(89.3,91.5) 


(92.1,93.9) 


Age 65-74 


89.6 


92.2 




(88.1,91.1) 


(90.9, 93.5) 


Age 75-84 


91.4 


93.8 




(89.8, 93.0) 


(92.4, 95.2) 


Age 85 and older 


90.9 


94.9 




(87.3, 94.5) 


(97.0, 100.0) 



A very high proportion of patients in the HCD had a follow-up visit after receiving 
a prescription measured at both six and eight months. In addition, if telephone consultations 
are included, 98.2 percent of patients had follow-up within eight months of receiving the 
prescription. This indicator measures follow-up for those receiving a prescription, not for 
those actually having a prescription filled. Conceptually, it seems desirable to include all 
those receiving a prescription in the denominator, as we have done, since this is the indicator 
of those who need follow-up. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-33 

cri mson\fi nal\ch ap3 . wpd\dpb 



Chapter 3 Results 

3.2.7 Anti-Depressant Follow-Up Rate 

Definition 

Percentage of beneficiaries with a prescription for a tricyclic or serotonin reuptake 
inhibitor with at least one outpatient visit to a primary care provider or mental health during 
the eight months after the prescription was written. 

Data Specifications 

Denominator : Aged beneficiaries enrolled in the HCD for at least eight months following 
the date a prescription was written for tricyclic or serotonin reuptake inhibitor. 
Numerator : Those with at least one visit to internal medicine or mental health during the 
eight months after receiving the prescription. 

HPHC Indicator Construction 

HCD: The AMRS was searched for patients with an appropriate prescription code 
during the period January 1, 1993 though April 30, 1995. We then subset to patients for 
whom we had data for at least 8 months following the prescription. This yielded a sample 
of 1,121 with an anti-depressant prescription. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-34 

crimson\fmal\chap3 wpd\dpb 



Chapter 3 



Results 



Results 



Below we present the rate of follow-up within 8 months after receiving a prescription 
for an anti-depressant. We also present the rate of follow-up after 6 months to determine the 
effect of varying the length of the episode. 



ANTI-DEPRESSANT 






FOLLOW-UP RATE 


Health Centers Division 




6 Months 


8 Months 


All eligibles 


90.8% 


93.2% 




(89.1,92.5) 


(91.7,94.7) 


Age 65-74 


92.6 


94.4 




(90.5, 94.7) 


(92.5, 96.3) 


Age 75-84 


88.0 


91.2 




(84.7,91.3) 


(88.8, 94.4) 


Age 85 and older 


91.1 


91.9 




(84.5, 97.7) 


(85.7, 98.3) 



The follow-up rate for anti-depressants in the HCD is almost identical to that for 
patients receiving an anti-hypertensive prescription. Additionally, if telephone consultations 
are included along with the face-to-face visits, 97.7 percent of patients receive follow-up care 
within eight months after receiving the prescription. As with anti-hypertensives, this 
indicator measures those receiving the prescription, not those having it filled. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-35 



Cha P ter3 Results 

3.3 Diagnosis Specific Care 

3.3.1 Rate of Post-hospital Follow-up for Myocardial Infarction 

Definition 

Percentage of beneficiaries hospitalized for MI with at least one primary care or 
cardiology visit within 60 days of discharge. 

Data Specifications 

Denominator: All aged beneficiaries continuously enrolled for at least 2 months in the study 

period following discharge for MI (ICD-9 = 410). 

Numerator: Those with one or more visits with a primary care provider or cardiologist 

within 60 days of discharge. 

HPHC Indicator Construction 

Hospitalization files were searched for the first non-transfer discharge with an 
appropriate ICD-9 diagnosis code for each beneficiary during each calendar year. We then 
subset to patients who were alive and for whom we had data for at least 60 days following 
discharge. This yielded samples of 32 and 36 patients for the two years in the MGD, and 84 
and 78 patients for the two years in the HCD. 

HCD: The AMRS was searched for encounters indicating the beneficiary had a visit 
with internal medicine or cardiology in the 60 days following discharge. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-36 

crimson\final\chap3.wpd\dpb 



Cha P ter3 Results 

MGD: Specialty is less well defined in the MGD than the HCD, due to the use of 
an Institutional -Interdivisional Care code that does not specify the specialty of the provider 
seen. We included visits with this code in determining whether the beneficiary had a visit 
within 60 days following discharge. 

Fee-for-Service Indicator Construction 

The MedPAR file was searched for the first non-transfer discharge for each 
beneficiary during each calendar year. The file was then subset to beneficiaries who were 
alive for at least 60 days in our sample following this discharge, yielding a sample of 2,994 
for 1994 and 2,948 for 1995. Claims were searched for an outpatient or office visit with a 
specialty coding of internal medicine, cardiology, or multispecialty clinic or group practice 
during the 60 day period. 

Results 

The rates of follow-up within 60 days after discharge from a hospitalization for a 
myocardial infarction were calculated for each year and averaged. The results were as 
follows: 



Health Economics Research, Inc. Access in Managed Care Plans: 3-37 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



MI FOLLOW-UP RATE 

All eligibles (age adjusted) 

Age 65-74 
Age 75-84 

Age 85 and older 



HPHC 



Fee-for-Service 


Division 


73.3% 


90.7% 


(71.7,74.9) 


(83.8, 97.6) 


77.6 


94.3 


(75.3, 79.9) 


(86.2, 100) 


73.4 


92.3 


(70.9, 75.9) 


(81.8, 100) 


60.2 


75.0 


(55.3,65.1) 


(38.8, 100) 



Health Centers Medical Groups 
Division 



93.2% 

(83.8, 100) 

94.9 

(82.8, 100) 

89.4 

(66.1, 100) 

100 

(83.5, 100) 



Both the HCD and the MGD had very small samples for this indicator. The 
proportion of the sample meeting the criterion to appear in our denominator (8 per thousand) 
is very consistent across the three groups and across the two years of data. This rate is 
consistent with findings by Hurst (1994) that the incidence rate of MI is 1 1 per thousand 
among the elderly, while 17 percent of those admitted die in the hospital (Federal Register, 
1994) and roughly 25 percent of those admitted die within 90 days of discharge (Dayhoff and 
Cromwell, 1994). Given the small samples (and the resultant wide confidence intervals), it 
is difficult to draw any conclusions regarding performance in the two divisions of the HMO. 



Health Economics Research, Inc. 

crimson\final\chap3 . wpd\dpb 



Access in Managed Care Plans: 3-38 



Chapter 3 Results 

3.3.2 Rate of Post-hospital Follow-up for Depression 

Definition 

Percentage of beneficiaries hospitalized for depression with at least one primary care 
or mental health visit within 14 days of discharge. 

Data Specifications 

Denominator : All aged beneficiaries continuously enrolled for at least 3 months in the study 

period follow discharge for depression (ICD-9 = 296, 298.0, 300.4, 301.12, 309.0, 309.1, 

311). 

Numerator : Those with one or more visits with a primary care or mental health provider 

within 14 days of discharge. 

HPHC Indicator Construction 

HCD: Institutional files were searched for the first non- transfer discharge with an 
appropriate ICD-9 diagnostic code for each beneficiary during each calendar year. We then 
determined whether the individual had another hospitalization for depression within 14 days 
following this discharge. Those with another such hospitalization during this period were 
dropped from the sample; those without a subsequent hospitalization formed the denominator 
for the indicator. This resulted in samples of 12 and 17 for HCD for 1994 and 1995. The 
AMRS was then searched for a record indicating the patient had a visit with an internal 
medicine or mental health specialist during the 14 days following discharge. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-39 

crimson\final\chap3 wpd\dpb 



Cha P ter3 . Results 

MGD: Hospitalizations were identified in a manner analogous to the HCD, yielding 
samples of 12 and 5 for MGD for the two years. Outpatient visits were then searched for 
evidence the person had a visit during the 14 days following discharge. As with other 
indicators based on specialty, the MGD claims suffer from the use of an Institutional- 
Interdivisional Care code that does not specify level of profession or specialty of provider l 
seen. We included visits with this code in constructing the indicator, but it doubtless is less 
specific than HCD. 

Fee-for-Service Indicator Construction 

Hospitalization files were searched for the first non-transfer discharge with an 
appropriate ICD-9 diagnostic code for each beneficiary during each calendar year. We then 
determined whether the individual had another hospitalization for depression within 14 days 
following this discharge. Those with another hospitalization were dropped from the sample. 
The resulting samples were 963 for 1994 and 895 for 1995. 

Results 

The rates of follow-up within 14 days after discharge from a hospitalization for 
depression were calculated for each year and averaged. The results were as follows: 



Health Economics Research, Inc. Access in Managed Care Plans: 3-40 

crimson\final\chap3 . wpd\dpb 



Cha P ter3 Results 



DEPRESSION FOLLOW- HPHC 

UP RATE 



Age 85 and older 



Fee-for-Service 


Division 


65.8% 


64.5% 


(62.7, 68.9) 


(37.3, 92.2) 


68.0 


79.1 


(63.1,72.9) 


(51.0, 100.0) 


64.8 


53.4 


(60.2, 69.4) 


(0, 100.0) 


63.0 




(53.5, 72.5) 





Health Centers Medical Groups 
Division 



All eligibles (age adjusted) 65.8% 64.5% 80.3% 

(52.6, 100.0) 

Age 65-74 68.0 79.1 87.5 

(61.6, 100.0) 

Age 75-84 64.8 53.4 75.0 



(16.5, 100.0) 



As with other indicators based on a hospitalization, creating the denominator for this 
indicator (persons hospitalized with a principal diagnosis of depression) is straightforward. 
In calculating the numerator, both fee-for-service and MGD data suffer from the coding of 
"groups" which contain unidentified or multiple specialties. Because of the prevalence of 
visits vvith this specialty code, and the appearance that much of what took place during these 
visits was appropriate follow-up care, we included them in the numerators of our measures. 
This practice would tend to bias these results upward relative to the HCD, in which specialty 
can be accurately defined and is narrowly limited to internal medicine and mental health 
specialists. A far more confounding problem in comparing the rates is the very small sample 
sizes for both divisions of the HMO which results in very imprecise estimates (note the wide 
confidence intervals for the HMO measures). The MGD had a total of 1 7 hospitalizations 
for depression across the 2 years and the HCD had 29. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-41 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

3.3.3 Rate of Follow-up for Abnormal Mammogram 

Definition 

Percentage of female beneficiaries with an abnormal mammogram who receive 
follow-up repeat mammogram, ultrasound, biopsy, or surgery within 15 days. 

Data Specifications 

Denominator: Female aged beneficiaries continuously enrolled for at least two months in 
the study period following an abnormal mammogram. 

Numerator : Those with repeat mammogram, ultrasound of breast, biopsy of breast lesion, or 
other surgical procedure of the breast within 15 days of abnormal result [AMRS = Y598 
(breast biopsy), Y215 (excision of breast lump), Y384 (excision of breast mass), T362 (fine 
needle aspiration: superficial tissue), T363 (fine needle aspiration: deep tissue), R200 (biopsy 
performed), T388 (aspirate cyst: breast), T391 (breast lump biopsy: needle directed), T392 
(breast lump biopsy: incisional), T393 (breast lump biopsy: excisional), R035 
(mammogram), R340 (mammogram-unilateral), R341 (mammogram-bilateral), R342 
(localize breast nodule or calcif. pre-op w/ marker), R261 (ultrasonography), TR188 
(ultrasound-breast), TR361 (cyst aspiration-ultrasound guidance), TR362 (needle biopsy- 
ultrasound guidance)]. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-42 

crimson\frnal\chap3.wpd\dpb 



Chapter 3 



Results 



HPHC Indicator Construction 

HCD: The Radiology Information System (RIS) was used to identify women with 
abnormal mammograms between June 1, 1994 and December 31, 1995. The RIS is a 
microcomputer-based dataset kept by the radiology department to keep track of all test results 
and to codify the readings of x-rays, i.e., normal, abnormal. The dataset was not created until l 
June 1994, so our sample is restricted to cases after that date. We then subset to those with 
at least two months of continuous enrollment after the date of the abnormal mammogram, 
yielding a sample of 76 women. The AMRS was then searched for follow-up procedure and 
test codes. 



Results 

The proportion of patients undergoing a follow-up within 15, 30, 45 and 60 days of 
the abnormal mammogram was as follows: 



ABNORMAL 




MAMMOGRAM 




FOLLOW-UP RATE 






Health Centers 




Division 


1 5 days 


46.1% 




(34.2, 58.0) 


30 days 


64.5 




(53.1,75.9) 


45 days 


77.6 




(67.6, 87.6) 


60 days 


90.8 




(84.0, 98.2) 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-43 



Chapter 3 Results 

3.4 Specialty Referral Care 

3.4.1 Population-Based Procedure Rates 

The construction and interpretation of our three population-based procedure 
rates — lens replacement, hip and knee replacement, and revascularization surgery — are all 
quite similar. Hence, we describe each of them in this section. 

Population-Based Rate of Lens Replacement 

Definition 

Rate of lens replacement per thousand beneficiaries during a 12 month period. 

Data Specifications 

Denominator : Aged beneficiaries continuously enrolled for at least 12 months during the 

study period. 

Numerator : The number of lens replacement surgeries (CPT = 66830-66986). 

Population-Based Rate of Hip and Knee Replacement 

Definition 

Rate of hip and knee replacement per thousand beneficiaries during a 12 month 
period. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-44 

crimson\fmal\chap3.wpd\dpb 



Chapter 3 Results 

Data Specifications 

Denominator : Aged beneficiaries continuously enrolled for at least 12 months during the 

study period. 

Numerator : Number of admissions for hip replacements (ICD-9 = 81.51, 81.53) and knee 

replacements (ICD-9 = 81.54, 81.55). 

Population-Based Rate of Coronary Revascularization 

Definition 

Rate of coronary revascularization procedures per thousand beneficiaries during a 12- 
month period. 

Data Specifications 

Denominator : All aged beneficiaries enrolled continuously for a twelve month period. 
Numerator : Admissions for a coronary revascularization procedure (bypass or angioplasty) 
( ICD-9 = 36.01, 36.02, 36.03, 36.04, 36.05, 36.09, 36.10-36.19, 36.2). 

HPHC Indicator Construction 

The MUPS enrollment file was used to identify all aged beneficiaries enrolled in the 
plan for either all of calendar year 1994 and/or all of calendar year 1995. This yielded a 
sample of 9,457 and 9,287 beneficiaries for the two years in the HCD, and a sample of 4,614 
and 4,454 beneficiaries for the two years in the MGD. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-45 

crimson\final\chap3.wpd\dpb 



Cha P ter 3 Results 

HCD: The institutional file was searched for claims with an appropriate CPT-4 code. 
(Data on all care outside the Health Centers is found in the institutional file.) We then 
determined the proportion of beneficiaries receiving one of the designated procedures. An 
individual was coded either as "0" not receiving a procedure or "1" receiving a procedure 
(beneficiaries receiving two lens replacements are coded the same as those receiving only v 
one). 

MGD: The institutional file was searched for claims with an appropriate CPT-4 code 
in a manner analogous to that used in the HCD. 

Fee-for-Service Indicator Construction 

To parallel the approach taken by HPHC, we identified all beneficiaries meeting the 
sample eligibility criteria for all of calendar year 1994 and/or for all of calendar year 1995, 
yielding samples of 363,934 and 329,1 16 beneficiaries for the two years. 

Since almost all lens replacements are performed on an outpatient basis, these 
operations were identified in the fee-for-service data using physician claims. The - 
physician/supplier file was searched for claims indicating a lens replacement. Modifiers and 
type of provider were then used to identify the actual surgeon's claims (as opposed to claims 
for pre- and post- operative care, or a surgical facility bill). 

For hip and knee replacement and revascularization rates, rates were constructed in 
a manner analogous to that used by HPHC. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-46 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



Results 

Procedure rates per one thousand eligibles, averaged across 1994-95, are presented 
below for each of our procedure groups. 



LENS REPLACEMENT 






HPHC 




Health Centers 


Medical Groups 




Fee-for-Service 


Division 






Division 


All eligibles (age adjusted) 


37.9 


32.5 






16.6 




(37.3, 38.5) 


(29.1,35.9) 






(12.6, 20.6) 


Age 65-74 


27.1 


19.4 






10.3 




(26.3, 27.9) 


(15.8,23.0) 






(6.3, 14.3) 


Age 75-84 


49.9 


39.6 






24.1 




(48.7,51.1) 


(32.1,47.1) 






(15.7,32.5) 


Age 85 and older 


44.6 


66.1 






39.2 




(42.5, 46.7) 


(39.2, 93.0) 






(15.0,63.4) 



HIP AND KNEE 
REPLACEMENT 


Fee-for-Service 




HPHC 




Health Centers 
Division 


Medical Groups 
Division 


All eligibles (age adjusted) 


6.8 

(6.5,7.1) 


5.9 

(4.3, 7.5) 






7.7 

(4.9, 10.5) 


Age 65-74 


7.0 
(6.6, 7.4) 


5.4 

(3.3,7.5) 






7.5 
(4.0, 13.0) 


Age 75-84 


7.3 
(6.8, 7.8) 


6.7 

(3.5, 9.9) 






8.6 

(3.4, 13.8) 


Age 85 and older 


3.4 
(2.8, 4.0) 


5.4 
(0, 14.1) 






5.4 
(0, 15.6) 



Health Economics Research, Inc. 

crimson\final\chap3 wpd\dpb 



Access in Managed Care Plans: 3-47 



Chapter 3 Results 



REVASCULARIZATION 
PROCEDURES 


Fee-for-Service 




HPHC 




Health Centers 
Division 


Medical Groups 
Division 


All eligibles (age adjusted) 


8.6 

(8.3, 8.9) 


7.6 

(5.7, 9.5) 






4.1 

(1.7,6.5) 


Age 65-74 


10.7 
(10.2, 11.2) 


7.7 
(5.4, 10.0) 






6.2 

(3.0, 9.4) 


Age 75-84 


7.8 

(7.3, 8.3) 


9.2 

(5.5, 12.9) 






2.5 

(0, 5.5) 


Age 85 and older 


1.8 

(1.4,2.2) 


1.3 

(0, 5.4) 






0.0 

(0,2.1) 



We would expect each of our procedures to be accurately coded in each of the three 
data sets, given their relatively expensive, "major procedure" nature. However, the small 
sample sizes for the HCD and MGD make it difficult to interpret the differences in these 
rates. Interpretation is also confounded by lack of a clear pattern — for instance the MGD has 
the lowest rate of revascularization procedures, but the highest rate of hip and knee 
replacement. In addition, even where procedure rates were significantly different across the 
three groups, with no appropriate benchmark it is impossible to determine whether a 
difference indicated overutilization in one sector or underutilization in another. 

3.4.2 Rate of Breast Cancer Oncology Follow-Up 

Definition 

Percentage of female beneficiaries with at least one oncology or general surgery visit 
in the six months following an initial diagnosis of breast cancer. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-48 

crimson\final\chap3 . wpd\dpb 



Cha P ter3 , Results 

Data Specifications 

Denominator: All female beneficiaries continuously enrolled for at least six months 

following an initial diagnosis of breast cancer. 

Numerator: Those with a first diagnosis of breast cancer (DH101, DH102) during the study 

period. 

HPHC Indicator Construction 

HCD: The AMRS has a function that allows it to search a medical record and 
indicate which occurrence of a given diagnosis code is the first occurrence. This feature was 
used to identify women whose first appearance of a breast cancer diagnosis code occurred 
during the period January 1, 1993 to December 31, 1995. This sample was subset to those 
with six months of continuous enrollment after the date of diagnosis. This yielded a sample 
of 162 women. 

Results 

The proportion of women who had a follow-up oncology or surgery visit during the 
six months after a breast cancer diagnosis was: 



Health Economics Research, Inc. Access in Managed Care Plans: 3-49 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



BREAST CANCER 
FOLLOW-UP RATE 






Health Centers 
Division 


All eligibles 


71.0% 

(63.7, 78.3) 


Age 65-74 


76.7 

(67.4, 86.0) 


Age 75-84 


70.9 

(58.0, 83.8) 


Age 85 and older 


41.2 

(14.9, 67.5) 



Identification of "first mention" of diagnosis in AMRS does not necessarily identify 
initial diagnosis of breast cancer, but rather the first entry of the code for breast cancer into 
the system. Therefore, it can reflect not only incident disease, but also the first entry in a 
record for a new member with a history of breast cancer in years past. For example, if a 75 
year old woman joined the HCD and reported during her initial visit that she had been treated 
for breast cancer 1 years earlier, a breast cancer diagnosis would be entered into the data. 
The search algorithm would identify the first occurrence of the diagnosis, but treatment for 
the condition would not necessarily be required. Given that 29 percent of women have no 
oncology/surgery visit, and 22 percent have no ambulatory encounters in the following six 
months despite continuous enrollment, this would merit further investigation. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-50 



Chapter 3 Results 

3.5 Primary Care 

3.5.1 Rate of New Enrollees with a Visit 

Definition 

Percentage of new enrollees with at least one visit during the first two months of 
enrollment. 

Data Specifications 

Denominator : All newly enrolled aged beneficiaries with at least three months of continuous 

enrollment in the study period. 

Numerator : Those with at least one face to face visit during the first two months of 

enrollment. 

HPHC Indicator Construction 

For each division, a sample was drawn of 750 new members who joined at age 65 or 
over in 1995 and were subsequently enrolled for at least 3 months. 

HCD: The AMRS was searched for a face to face visit during the first 60 days of 
membership. 

MGD: Claims were searched for a CPT code indicating an evaluation and 
management visit during the first 60 days of membership. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-51 

crimson\final\chap3 . wpd\dpb 



Chapter 3 



Results 








The proportion of new 


enrollees with a visit within 60 days was as 


follows: 




RATE OF NEW 
ENROLLEES WITH A 
VISIT 


Health Centers 
Division 


Medical Groups 
Division 






All eligibles 


73.9% 

(70.7,77.1) 


48.7% 

(45.1,52.3) 






Age 65-74 


74.6 

(70.5, 78.7) 


44.1 

(39.3, 48.9) 






Age 75-84 


72.8 

(66.7, 78.9) 


53.6 
(47.2, 60.0) 






Age 85 and older 


72.6 
(60.7, 84.5) 


61.3 

(48.4, 74.2) 





The proportion of new enrollees with a visit in 60 days is substantially higher in the 
HCD than the MGD (although the differences narrow among the older age groupings). To 
determine whether the difference between the divisions disappeared with a larger window, 
we also determined the proportion of beneficiaries with a visit at 30 day intervals up to 1 80 
days. These results are presented in Figure 3-1. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-52 



Chapter 3 



Results 



Figure 3-1 

Proportion of New Enroilees With a Visit at 30 Day 
Intervals After Enrollment 



• — « 

> 

0* 



VI 

a 



VI 

5 

"o 
i_ 

c 

W 

Z 

V. 

O 

e 
o 
- S 
t 

o 
c 
o 

L. 

0. 




60 90 120 

Days After Enrollment 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-53 



Chapter 3 



Results 



RATE OF NEW 






ENROLLEES WITH 






A VISIT 








Health Centers 


Medical Groups 




Division 


Division 


30 Days 


52.4% 


34.0% 


60 Days 


73.9% 


48.7% 


90 Days 


83.5% 


57.6% 


120 Days 


86.9% 


65.4% 


150 Days 


89.4% 


70.8% 


180 Days 


91.4% 


74.7% 



Two factors likely contribute to the higher visit rate for HCD at every period. First, HCD 
has implemented a more comprehensive system for screening and intake of new Medicare 
beneficiaries to ensure that these members are seen in a timely manner. Second, MGD 
members (unlike the HCD) may be changing insurers (for example, moving from fee-for- 
service to HPHC) but remaining with the same physician. Thus, although new HPHC - 
enrollees, they would not be new to the group practice, and would not need an "initial" 
evaluation examination. Without additional case-study work or beneficiary interviews, it is 
impossible to determine which is the more prevalent factor. 



Health Economics Research, Inc. 

crimson\final\chap3 wpd\dpb 



Access in Managed Care Plans: 3-54 



Chapter 3 Results 

3.5.2 Rate of Beneficiaries with a Visit 

Definition 

Percentage of beneficiaries with at least one visit with a primary care physician or 
specialist during a 1 2 month period. 

Data Specifications 

Denominator : All aged beneficiaries continuously enrolled for at least 12 months in the study 

period. 

Numerator : Those with at least one visit to a primary care physician or specialist, excluding 

optometry/ophthalmology. 6 

HPHC Indicator Construction 

For each division of HPHC, a random sample of 500 beneficiaries was drawn from 
those continuously enrolled for all of 1995 using the enrollment file. 

HCD: The AMRS was searched for encounters indicating the beneficiary had a visit 
during the calendar year. 

MGD: Claims were searched for CPT-4 codes indicating the beneficiary had an 
evaluation and management visit during the calendar year 



6 Routine eye exams for prescribing glasses are not covered under fee-for-service Medicare. Hence, we exclude this 
specialty from the analysis since the managed care and fee-for-service benefits are very different. 

Health Economics Research, Inc. Access in Managed Care Plans: 3-55 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



Fee-for- Service Indicator Construction 

The denominator file was used to identify beneficiaries who were enrolled for all of 
calendar year 1995. This yielded a sample of 325,984 beneficiaries. Physician/supplier 
records were then searched for a CPT-4 code indicating the beneficiary had an evaluation and 
management visit during the calendar year. 



Result 

For each group, we calculated the proportion of beneficiaries with at least one visit 
during the calendar year. The results are as follows: 



RATE OF 








BENEFICIARIES WITH 








A VISIT 












Health Centers 


Medical Groups 




Fee-for- 


Division 


Division 




Service 






All eligibles (age 


88.4% 


93.9% 


90.9% 


adjusted) 


(88.3, 88.5) 


(91.7,96.1) 


(88.3, 93.5) 


Age 65-74 


86.0 


95.0 


89.7 




(85.8, 86.2) 


(92.3, 97.7) 


(85.8, 93.6) 


Age 75-84 


90.6 


91.6 


92.7 




(90.4, 90.8) 


(87.2, 96.0) 


(88.6, 96.8) 


Age 85 and older 


91.4 


97.6 


90.0 




(91.1,91.7) 


(91.7, 100.0) 


(81.6,98.4) 



Annual visit rates for fee-for-service and the two divisions of HPHC are all quite 
similar. All are also noticeably higher than the 76.9 percent visit rate found by PPRC (1 995). 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-56 



Cha P ter3 . Results 

The difference may result form PPRC's use of a national sample, while our fee-for-service 
sample is primarily in the Boston metropolitan area. 

3.5.3 Continuity of Care Index 

Definition 

Proportion of visits per patient for primary care that are with the patient's primary 
care physician. 

Data Specifications 

Denominator: All visits for primary care for aged beneficiaries continuously enrolled for at 

least 12 months in the study period. 

Numerator : Visits with the patient's primary care physician. 

HPHC Indicator Construction 

HCD: A sample of 500 aged beneficiaries continuously enrolled for all of 1995 was 
randomly selected from the enrollment file. Then, the AMRS was used to identify visits with 
internal medicine during the year, yielding a sample of 464 patients with at least one visit, 
and a sample of 393 patients with more than one visit. (Since patients with one visit, by 
definition have a continuity index of 100%, only those two or more visits were used for 
constructing the indicator.) For each beneficiary, the most frequent provider was determined 
using the provider code and that individual was designated as the primary care provider. The 



Health Economics Research, Inc. Access in Managed Care Plans: 3-57 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



count of number of visits with that provider and total number of visits to internal medicine 
was constructed for each beneficiary. 

Results 

We calculated the proportion of primary care visits each beneficiary had with the 
primary care physician. The average proportions were: 



CONTINUITY OF CARE INDEX 




Health Centers 




Division 


All eligibles 


71.3% 


Age 65-74 


69.1 


Age 75-84 


73.5 


Age 85 and older 


76.2 



The average proportion of internal medicine visits with the primary care provider (for 
eligibles having more than one visit) was just over 70 percent. This proportion did not vary 
substantially across our three age groups. This indicator was constructed only for the HCD, 
since the specialty coding is far superior in these data sets than in the MGD or fee-for-service 
data. In the HCD, "primary care" visits can be identified much more accurately than in the 
other sectors. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-58 



Chapter 3 Results 

3.6 Analysis of Disenrollees 

The measures reported above capture utilization experience for patients with a wide 
range of diagnoses receiving a wide range of services. However, as is necessary in any 
limited list of indicators, we cannot cover all conditions for which patients might seek care 
or treatments they might receive. Thus, it is desirable to also develop some broader 
indicators of patient's experiences in the health care system, that, although less firmly rooted 
in clinical standards of care, may reflect whether patients feel they are receiving "timely and 
appropriate" care. 

Unlike most managed care enrollees, who may be quite restricted in their ability to 
leave an HMO and acquire other health insurance, Medicare enrollees can switch to fee-for- 
service (or another managed care plan) with only 30 days notice. Thus, the characteristics 
of beneficiaries who disenroll, and their experience both while in the health plan and 
immediately after leaving, may provide some evidence of dissatisfaction with the care being 
received. For example, patients with chronic high cost conditions may disenroll as a 
response to perceived barriers to care. High levels of utilization soon after enrollment in the - 
fee-for-service system may reflect "pent-up demand," especially for high-cost procedures or 
specialty care. 

To explore these issues, our analysis of disenrollees contains three components: 
(1) calculation of disenrollment rates, (2) comparison of characteristics of disenrollees with 
enrollees, and (3) disenrollees' patterns of fee-for-service care after leaving the HMO. 
Although our analysis of components (1) and (2) comes from HPHC data, an alternative for 



Health Economics Research, Inc. Access in Managed Care Plans: 3-59 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

calculating disenrollment rates and comparing demographic and enrollment charactistics of 
disenrollees and enrollees would be to use enrollment data maintained by HCFA. 
Comparison of utilization while in the HMO requires use of the HMO data files. 

3.6.1 Disenrollment Rates l 

Definition 

Percentage of beneficiaries disenrolling from HPHC (excluding those who died) 
during the calendar year. 

Data Specifications 

Denominator : Aged beneficiaries enrolled in HPHC for any part of the calendar year. 

Numerator : Those disenrolling from HPHC (excluding those who died). 

Ideally, we would like to measure the number of "voluntary" disenrollees, excluding 
those who "involuntarily" disenrolled due to death or relocation. Although we have no - 
information on relocation, HPHC does have an accurate count of disenrollees who died. 7 
Thus, deaths are excluded from the numerator. This approach differs from HEDIS 3.0, which 
counts deaths as disenrollees. This approach also differs from HEDIS in that we count 
everyone ever enrolled during the year in the denominator, rather than comparing 
enrollments at the endpoints of two years. Thus, our approach would count beneficiaries 



7 To check the validity of the HPHC death variable, we compared HPHC membership end date and death information 
against HCFA's denominator file. The HPHC death variable was found to be accurate for 98 percent of disenrollees. 

Health Economics Research, Inc. Access in Managed Care Plans: 3-60 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



who enrolled and left during the course of a year (for example, enrolled in April, left in 
August) in the numerator and denominator, while HEDIS would not. 

Results 

Disenrollment rates for the two divisions, calculated using HPHC administrative 
enrollment files for 1994 and 1995, are as follows: 



DISENROLLMENT RATE 








Health Centers 


Medical Groups 




Division 


Division 


1994 


3.2% 


3.4% 


1995 


2.8% 


3.8% 



These rates are substantially lower than the industry average of 9.2 percent per year 
(excluding deaths) reported in the Public Sector Contracting Report (1997), the 14 percent 
found by Riley, et al. (1997) for Medicare beneficiaries or the 17 percent rate reported by the 
GAO for selected markets (1996). However, it should be noted that disenrollment rates are 
sensitive to the definition of who was ever enrolled and who disenrolled. For example, if 
beneficiaries who cancelled applications before the effective enrollment date and retroactive 
disenrollment date included, the disenrollment rate will be higher that if these beneficiaries 
are excluded (GAO, 1996). 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-61 



Cha P ter3 . Results 

3.6.2 Comparison of Enrollees and Disenrollees 

Demographic Characteristics 

Below we compare the mean age and percentage who are female for those enrolled 8 
for any part of 1994 or 1995 with those disenrolling during those two years. 





Health Centers Division 


Medical Groups Division 




Ever Enrolled 


Disenrolled 


Ever Enrolled 


Disenrolled 


Sample Size 


12,838 


664 


8,909 


492 


Percent Female 


59% 


57% 


57% 


62% 


Mean Age 


72 


73* 


72 


74* 



The proportion of females disenrolling did not differ significantly from the overall HPHC 
Medicare enrollment for either division. Disenrollees were significantly older than the 
average Medicare enrollee (as designated by the asterisk), but only by one or two years on 
average. 

Length of Enrollment 

For each of the two divisions of HPHC, we calculated (a) mean length of enrollment 
for those disenrolling (and not believed to be dead), and (b) mean length of enrollment for 
those enrolled as of the end of the calendar year. These results are presented in Table 3-2. 

For three of the four groups, disenrollees had significantly shorter lengths of 
membership in HPHC than did those enrolled at the end of the year. However, average 



"Those ever enrolled includes the sample that eventually disenrolled. 



Health Economics Research, Inc. 

criffison\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-62 



Chapter 3 



Results 



Table 3-2 



Mean Length of Enrollment, For Disenrollees and Those Enrolled 
at the End of the Calendar Year 



Mean Length of 



HCD 1994 

Enrolled 

Disenrolled 

MGD 1994 

Enrolled 

Disenrolled 

HCD 1995 

Enrolled 

Disenrolled 

MGD 1995 

Enrolled 

Disenrolled 



Sample Size 


Enrollment (months) 


10,064 


81.3 


333 


61.0 * 


4,923 


68.9 


174 


64.9* 


11,650 


77.6 


331 


52.9* 


8,132 


47.5 


318 


46.4 



* Indicates statistical difference at the 5 percent level. 

NOTE: The mean length of enrollment in 1995 is substantially lower than in 1994 for both divisions because 
of the large influx of new Medicare members during 1995. 

SOURCE: HPHC enrollment file. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-63 



Chapter 3 Results 

length of membership is quite long for each of the disenrollee groups, ranging roughly from 
4 to 6 years. 

Disenrollee Care while in HPHC 

For disenrollees who had been members of the plan for at least one year prior to * 
leaving (and who did not die), we extracted data on outpatient utilization for the twelve 
months prior to disenrolling. Disenrollees were pooled for the two years 1 994 and 1 995 to 
increase sample sizes. For comparison, we selected random samples of beneficiaries 
enrolled for all of 1995 and extracted their outpatient utilization as well. 

Table 3-3 presents comparisons of enrollee and disenrollee inpatient utilization. 
Disenrollees are significantly more likely to have been hospitalized during the 12 months 
than those who remain enrolled, both in the HCD and the MGD. Among those with a 
hospitalization, disenrollees had a higher mean number of stays in both divisions. 

Similar data on outpatient utilization is reported in Table 3-4. In both the HCD and 
MGD, enrollees and disenrollees were very likely to have an outpatient contact during the - 
12 month period. In the HCD, enrollees were slightly (and significantly) more likely to have 
had a face-to-face visit (95.4%) than were disenrollees (91.9%). In both divisions, 
disenrollees are lower utilizers of outpatient physician services, with fewer mean, median, 
and tenth percentile values for each type of contact than those still enrolled. 

These results are difficult to interpret, given the conflicting results for inpatient and 
outpatient utilization. The higher rates of hospitalizations are consistent with the results of 
Morgan et al (1997), who found that sicker beneficiaries are more likely to disenroll than 

Health Economics Research, Inc. Access in Managed Care Plans: 3-64 

crimson\finaI\chap3.wpd\dpb 



Chapter 3 



Results 



Table 3-3 



Inpatient Utilization for Disenrollees and Enrollees During a 12 Month 
Period During Which They Were Enrolled in HPHC 



Percent with a Hospitalization 
Mean Number of Hospitalizations 
(For those Hospitalized) 



Health Centers Division 



Enrollees 

13.0 % 
1.3 



Disenrollees 

17.7 % * 
3.2 * 



Percent with a Hospitalization 
Mean Number of Hospitalizations 
(For those Hospitalized) 



Medical Groups Division 



Enrollees 

12.6 % 
1.2 



Disenrollees 

21.5 % * 
1.5 * 



NOTE: Data for disenrollees covers the 12-month period prior to disenrollment from HPHC. 
Data for enrollees covers a 12-month enrollment period in HPHC. 

* Indicates statistical difference at the 5 percent level. 

SOURCE: HPHC inpatient files. 



Health Economics Research, Inc. 

crimson\fmal\chap3 . wpd\dpb 



Access in Managed Care Plans: 3-65 



Chapter 3 



Results 



Table 3-4 



Outpatient Utilization for Disenrollees and Enrollees During a 12 Month Period 
During Which They Were Enrolled in HPHC 



Health Centers Division 



Enrollees 



Disenrollees 



Percent with a Visit 
Mean Number of Visits 
Median Number of Visits 
Tenth percentile Number of Visits 



95.4 % 

12 

9 

3 



91.9%* 
9 * 

7 
1 



Medical Groups Division 



Percent with a Claim 
Mean Number of Claims 
Median Number of Claims 
Tenth percentile Number of Claims 



Enrollees 
94.6 % 
36 

23 
6 



Disenrollees 
94.4 % 
15 * 

10 

3 



NOTE: Data for disenrollees covers the 12-month period prior to disenrollment from HPHC. 
Data for enrollees covers a 12-month enrollment period in HPHC. 



* Indicates statistical difference at the 5 percent level. 
SOURCE: HPHC encounter and claims outpatient files. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-66 



Chapter 3 Results 

their healthier counterparts. Given this result, we would have expected to also see higher 
outpatient utilization for disenrollees. The disenrollees' lower outpatient utilization could 
be indicative of problems accessing outpatient care or could be a function of high-utilizers 
self-selection into (and continued enrollment in) managed care because of the convenience 
and low out-of-pocket expenses. 

3.6.3 Disenrollee Care after Leaving HPHC 

In addition to the analysis of experience while in the plan, we were also interested in 
care received by disenrollees after leaving HPHC. For this analysis, we constructed two 
comparison groups: (1) beneficiaries never in managed care during our study period, (2) 
beneficiaries in fee-for-service who eventually enrolled in managed care. The first group 
represent the "typical" beneficiaries, most of whom remain in fee-for-service. The second 
group may be more similar to disenrollees, who at one point thought they would prefer 
managed care. 
Sample Selection 

Two files maintained by HPHC were used in creating the sample of beneficiaries for 
the analysis of disenrollee care after leaving the HMO. One file contained a "Plan Record 
Number" along with member date of birth, sex, date membership ended and whether or not 
the member was believed to have disenrolled because of death. A second file contained the 
member's Medicare HICNO. These two files were merged, and the resulting file was 



Health Economics Research, Inc. Access in Managed Care Plans: 3-67 

crimson\final\chap3 . wpd\dpb 



Cha P ter3 Results 

matched against HCFA's denominator file which provides demographic and enrollment data 
on Medicare beneficiaries. 

Of the 901 beneficiaries disenrolling during the period December 31, 1993 to 
September 30, 1995 who were believed to still be alive by HPHC, 84 percent (758 
beneficiaries ) were matched to data on the denominator file. (Disenrollees that did not 
match to the denominator most likely moved out of our study area, which covered 
Massachusetts, New Hampshire, Rhode Island, and Vermont.) 9 We then subset to members 
for whom we had at least 3 months of fee-for-service data in our study period, yielding a 
sample of 373 disenrollees. (The demographic distribution of these disenrollees is shown 
in table 3-5). Of the remaining disenrollees, 364 were in managed care for at least one month 
of the three following disenrollment, and 21 died within 3 months of disenrolling. (Of those 
in managed care, two re-enrolled in HPHC the day after initial disenrollment— no others 
returned to this plan.) 

Two randomly selected comparison groups of 1,000 beneficiaries each were then 
drawn from the denominator file. The first consisted of 1,000 people who were aged 
Medicare beneficiaries during the entire 1994-95 period, but did not belong to a managed 
care group at any time during that period. The second consisted of 1,000 people who were 
aged Medicare beneficiaries who joined an HMO for whom we had at least 3 months of fee- 
for-service data in our study period before they entered managed care. All members of the 



'Although the denominator file contains a 100 percent sample of beneficiaries, the "finder file" used in identifying 
beneficiaries for our analysis contained only these states. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-68 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

comparison group were required to reside in the HPHC catchment area, defined by zip codes, 
as of January 1994. 

Because of concerns that medical utilization might vary seasonally, we determined 
from the HPHC sample the proportion of members disenrolling during each month of the 
study, and drew our comparison samples accordingly. That is, we drew a total of 22 samples 
for each comparison group, corresponding to disenrollees from the plan for the 22 months 
from the end of December, 1993 to the end of September, 1995. (The 22 samples contained 
a total of 1,000 beneficiaries.) For example, 3.7 percent of the HPHC disenrollees left the 
plan December 31,1 993 . The follow-up period for these beneficiaries thus becomes January 
- March, 1994. We then drew data for 3.7 percent of the "never in managed care group" 
during the period January - March, 1994. (The 37 beneficiaries were selected randomly.) 
For the "about to join managed care group" we randomly selected 37 beneficiaries who 
joined managed care in April 1994, thus making their three months in the study January - 
March, 1994. Beneficiaries in the comparison groups were drawn without replacement so 
the same individual could not be in the sample twice. 

Table 3-5 compares the age and gender distributions for disenrollees to the two 
comparison groups. The age distribution for the never in managed care group differs 
significantly from the HPHC disenrollees, with more disenrollees being in the younger age 
groupings. The age distribution for those about to enter managed care does not differ from 
the HPHC disenrollees. Neither group differs from the disenrollees in terms of gender 
distribution. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-69 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



Table 3-5 



Samples for Disenrollment Analysis 







Those Never 


Those about 




HPHC 


Enrolled in 


to Enroll in 




Disenrollees 


Managed Care 


Managed Care 


Sample Size 


373 


1,000 


1,000 


Age 65-74 


54.4 % 


42.0 % 


52.1 % 


Age 75-84 


34.3 


42.9 


39.4 


Age 85 and older 


11.2 


15.1 


8.5 


Gender 








Female 


54.9 


60.1 


58.3 


Male 


45.1 


39.9 


41.7 



NOTE: The age distribution of those never enrolled in managed care is statistically different than that of the HPHC 
Disenrollees at the 5% level. Distributions by gender are not significantly different. 

SOURCE: HPHC Enrollment File; random sample drawn from HCFA's Denominator file. 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-70 



Chapter 3 Results 

Utilization Measures 

Given the relatively small sample of disenrollees, we did not have the statistical 
power to look for differences in utilization for individual services. For example, we did not 
try to determine whether the rates of elective procedures such as cataract surgery or major 
joint replacement were different across the three analytic samples. Instead, we used broad i 
categories of utilization. For inpatient care, we examined the proportion of the sample with 
a hospitalization, mean number of hospitalizations (for those with at least one), Medicare 
Part A payments per user, and Medicare Part A payments per eligible. 

For physician/supplier care, we aggregated claims using BETOS groupings (Berenson 
and Holahan, 1990) into six classes: all physician/supplier services, visits (excluding mental 
health), mental health visits, procedures, imaging services, and tests. We then calculated 
average allowed charges per beneficiary, proportion of beneficiaries with positive allowed 
charges, and average allowed charges per user. Disaggregating average allowed charges per 
beneficiary into its two components allows us to investigate whether differences arise from 
differences in the number of beneficiaries receiving the service, or from differences in the 
intensity of services received. (Allowed charges, rather than a count such as number of visits 
or number of tests, were used as a measure of service intensity so that more costly, higher 
intensity services would be weighted more heavily than lower cost services. For example, 
one MRI counts much more heavily than one chest x-ray in the imaging category using 
allowed charges, whereas counts of procedures would weight both equally.) We chose a 
three month follow-up because we wanted a period sufficiently long for "pent up" demand 



Health Economics Research, Inc. Access in Managed Care Plans: 3-71 

crimson\final\chap3.wpd\dpb 



Chapter 3 Results 

to be observable but brief enough that unrelated new conditions developing after 
disenrollment would not be included. 

Results 

Table 3-6 presents results of hospital utilization for a three-month period for each of '" 
the three analytic groups. Payments per eligible and the proportion of eligibles with a 
hospitalization are slightly higher for those never in managed care than for HPHC 
disenrollees, who in turn have slightly higher values than those about to enter managed care. 
However, none of the differences in utilization measures are statistically significant across 
the three analytic groups. This result differs from that of Morgan et al (1997) who found that 
disenrollees had significantly greater inpatient utilization following disenrollment than fee- 
for-service Medicare beneficiaries. The difference may result from our smaller sample size 
and resulting lower statistical power. However, Morgan's sample, drawn from the southern 
Florida area which has a very high disenrollment rate (GAO, 1 996), may not generalize to 
all managed care organizations and to all markets. 

Table 3-7 presents results of physician/supplier utilization for a three-month period 
for the same three analytic groups. HPHC disenrollees have higher physician charges per 
eligible ($416.86) than those about to enroll in managed care ($219.59). The difference is 
statistically significant, as is the difference between number of eligibles receiving a physician 
supplier service (77.1% for HPHC disenrollees vs. 68.3% for those about to enroll) and 
allowed charges per user ($540.67 for HPHC disenrollees vs. $321.51 for those about to 



Health Economics Research, Inc. Access in Managed Care Plans: 3-72 

crimson\final\chap3.wpd\dpb 



Chapter 3 



Results 



Table 3-6 



Hospital Utilization for a 3 Month Period 







Comparison Group 




Those Never 


Those About 






Enrolled in 


to Enroll in 




HPHC Disenrollees 


Managed Care 


Managed Care i. 


Payments per Eligible 


$372.14 


$388.89 


$360.68 


Proportion of eligibles who are users 


4.3% 


5.0% 


2.9% 


Average hospitalizations per user 


1.2 


1.4 


1.3 


Payments per user 


$8,654.42 


$7,777.90 


$12,437.24 



NOTES: 

1) Results are age-sex adjusted. 

2) No statistically significant differences exist between HPHC disenrollees and the other groups. 

SOURCE: MedPAR files, 1994 and 1995. 



Health Economics Research, Inc. 

crimson\final\chap3 .wpd\dpb 



Access in Managed Care Plans: 3-73 



Chapter 3 



Table 3-7 



Results 



Part B Utilization During a 3 Month Period for Disenrollees, Those 
Always in Fee-For-Service, and Those About to Enroll in Managed Care 







Those Never 


Those About 




HPHC 


Enrolled in 


To Enroll in 




Disenrollees 


Managed Care 


Managed Care 


All Physician/Supplier Services 
Allowed charges per eligible 
Proportion of eligibles using service 
Allowed charges per user 


$416.86 
77.1% 
$540.67 


$349.21 
74.3% 
$470.06 


$219.59 ** 

68.3% * 
$321.51 * 


Phvsician Visits 
Allowed charges per eligible 
Proportion of eligibles using service 
Allowed charges per user 


$133.08 
71.7% 
$185.61 


$143.70 
70.4% 
$191.54 


$86.87 * 

62.9% * 

$138.11 * 


Procedures 
Allowed charges per eligible 
Proportion of eligibles using service 
Allowed charges per user 


$110.46 
23.9% 
$462.18 


$84.99 
26.1% 

$325.73 


$63.34 * 
18.9% 

$335.13 


Tests 
Allowed charges per eligible 
Proportion of eligibles using service 
Allowed charges per user 


$37.80 
46.8% 
$80.76 


$33.98 
44.9% 
$75.64 


$23.34 * 
34.3% * 
$68.06 


Imaging Studies 
Allowed charges per eligible 
Proportion of eligibles using service 
Allowed charges per user 


$35.35 
25.5% 
$138.64 


$31.58 
27.3% 
$115.78 


$25.87 
20.0% 
$129.37 


Mental Health Visits 
Allowed charges per eligible 
Proportion of eligibles using service 
Allowed charges per user 


$18.00 
6.8% 
$264.73 


$8.48 * 
3.3% * 
$257.34 


$2.13 * 

1.8% * 

$118.61 * 



NOTES: 

1) All results are age-sex standardized. 

2) * indicates statistical difference from disenrollees at the 5 percent level. 

S OURCE. Xati a wnl Claimo ll i it a nji Phjfririin/Siipplirr filrs 1994 nnd 1995 . 



Health Economics Research, Inc. 

crimson\final\chap3.wpd\dpb 



Access in Managed Care Plans: 3-74 

crimson\final\chap3tabs\Tab3-7\nd 



Chapter 3 Results 

enroll). Utilization measures for HPHC disenrollees were not significantly different from 
those never enrolled in managed care. 

The results for physician visits, procedures, and tests are similar, with HPHC 
disenrollees having significantly higher utilization than those about to enroll. There are no 
significant differences among the imaging studies category. 

The only category for which HPHC disenrollees have significantly higher utilization 
than both those never enrolled and those about to enroll is for mental health visits. Allowed 
charges per eligible are twice as high ($18.00 vs. $8.48) for the disenrollees as for those 
never enrolled, and more than five times as high ($18.00 vs. $2.13) for the disenrollees as 
for those about to enroll in managed care. The proportion using mental health services is 
also significantly higher for HPHC disenrollees than for either of the other groups, while 
allowed charges per user among HPHC disenrollees are more than twice as high as those for 
beneficiaries about to enroll in managed care. 

The comparisons between those never enrolled and those about to enroll are 

consistent with favorable selection bias among Medicare beneficiaries for managed care - 

organizations, with those younger and healthier being more likely to enroll (and stay 

enrolled). Comparisons between HPHC disenrollees and the two groups indicate that while 

a very small proportion of Medicare beneficiaries disenroll from HPHC, those who do have 

high rates of utilization after leaving the Plan. It is not clear, however, if this is the result of 

poorer health among those leaving or a pent-up demand for services that were not received 

while in the HMO. The most significant finding was the use of mental health care for 

disenrollees. Our case study (Appendix A) describes the mental health benefits and 

Health Economics Research, Inc. Access in Managed Care Plans: 3-75 

crimson\final\chap3 wpd\dpb 



Chapter 3 Results 

providers available to HPHC members. Unfortunately, we do not have comparable post- 
disenrollment utilization data for beneficiaries who switch to another managed care plan after 
lwaving HPHC. These beneficiaries account for nearly half the sample disenrolling during 
our study period. Results from Nelson el al (1997) suggest that switchers more closely . 
resemble HMO enrollees who stay in the managed care organization than those who return 
to fee-for-service. Nevertheless, the high-utilization for beneficiaries returning to fee-for- 
service is important in itself. 



Health Economics Research, Inc. Access in Managed Care Plans: 3-76 

crimson\final\chap3.wpd\dpb 



4 



Discussion 



The purpose of this study was two-fold. The first purpose was to develop a series of 
Medicare performance indicators that could be applied to both managed care and fee-for- 
service data. The second was to operationalize these indicators using Medicare fee-for- 
service and Medicare managed care data, to determine whether the indicators could in fact 
be implemented in a meaningful manner. In this chapter, we discuss the results and 
implications of the study. We begin by focusing narrowly on the results from the fee-for- 
service and managed care data and discuss the interpretation of our quantitative results. We 
then discuss more broadly the "lessons learned" from conducting the study and the 
implications of our findings for developing a performance monitoring system for Medicare 
managed care. 

4.1 Comparison of Fee-For-Service and Managed Care Results 

In this study, we compare performance in the fee-for-service sector with two divisions 
of Harvard Pilgrim Health Care (HPHC): the staff-model Health Centers Division (HCD) 
and the IPA-type Medical Groups Division (MGD). The two divisions have very different 
physician contracting and payment arrangements and different institutional structures, 



Health Economics Research, Inc. Access in Managed Care Plans: 4-1 

crimson\final\chap4.wpd\dpb 



Chapter 4 Discussion 

allowing us to compare results within the HMO as well as between managed care and fee-for- 
service. This section briefly summarizes the results for our 19 performance indicators. 

Preventive Care 

Given the HMO's incentives to contain costs of future care and the philosophical ■- 
emphasis on prevention, we expected that performance in the HMO would surpass that of 
fee-for-service practice. This was clearly the case for the colon cancer screening rate, which 
was over 50 percent for both divisions of the HMO, while only 36 percent of fee-for-service 
beneficiaries received any type of screening test during the 24-month study period. 1 Nearly 
twice as high a proportion (77 percent) of aged women in the HCD received breast cancer 
screening during the 24-month period compared with women in fee-for-service (40 percent);, 
performance in the MGD was between these two, with 67 percent of women receiving the 
test. We expect that much of the difference resulted from the use of an automated reminder 
system in the HCD that notifies physicians when a member is due for mammography. In this 
instance, the managed care "philosophical emphasis" on prevention and the financial 
incentives to provide preventive services have been institutionalized into a reminder system 
to help insure that services are in fact provided. In the MGD, which has no such automated 
system, methods of "reminding" physicians that care is due vary across the groups, and 
consequently the rate of mammography is lower. 



Fee-for-service coverage of fecal occult blood tests was limited during our study period, contributing to the low 
figure. 

Health Economics Research, Inc. Access in Managed Care Plans: 4-2 

crimson\final\chap4.wpd\dpb 



Cha P ter 4 Discussion 

Chronic Disease Care 

Care for chronic diseases is an area where managed care has the potential to 
outperform fee-for-service because of the greater ability (and incentives) to coordinate care 
and manage cases through a primary caregiver. HPHC has been in the process of developing 
automated reminders for specific conditions (such as diabetes) and guidelines for treatment l 
of common conditions (such as many of the ambulatory care sensitive (ACS) diagnoses). 
On the other hand, there are concerns that patients with chronic diseases, who may be quite 
expensive to treat, may be underserved and see their health deteriorate in managed care 
(Ware, et ah, 1996). The extent to which HMO initiatives to coordinate care will actually 
result in "care management" as opposed to cost reduction through "utilization management" 
has not been demonstrated. 

Our study found that both divisions of the HMO performed quite well in treating 
chronic conditions. Rates of secondary preventive services for diabetics were higher in the 
HMO than in fee-for-service, while the admission rates for ambulatory care sensitive 
conditions were lower (meaning that fewer patients reached the point which required a . 
hospitalization). 2 Rates of outpatient care pre- and post- ACS admission were quite high 
(80-85 percent) for both fee-for-service and managed care, indicating that most patients did 
have contact with the medical system before and after their actual hospitalization. For the 
HCD (which has computerized data on prescriptions), we also found that rates of follow-up 



2 

Admission rates for ACS conditions may reflect the health status of the study population. If HPHC beneficiaries are 
healthier (controlling for age) it would help explain differences in hospitalization rates. 

Health Economics Research, Inc. Access in Managed Care Plans: 4-3 

crimson\final\chap4.wpd\dpb 



Cha P ter4 Discussion 

for patients with prescriptions for anti-hypertensive or anti-depressant medications were quite 
high (over 90 percent). 

Dia gnosis Specific Care 

Our three indicators for diagnosis-specific care highlight the problem inherent in l 
developing this type of indicator. By focusing on a very specific condition (or incident) it is 
possible to develop an indicator for which there is a consensus on appropriate treatment. 
However, the narrow focus also implies that sample sizes quickly become an issue. 

The conditions we chose (myocardial infarction, hospitalization for depression, 
abnormal mammogram) are not rare or exotic conditions among the elderly. However, given 
the number of Medicare beneficiaries enrolled in HPHC, and the resulting small samples and 
wide confidence intervals, it is difficult to draw any conclusions regarding performance 
across the three sectors. 

Specialty Care 

Perhaps more than any other area, skeptics of managed care worry about the 
incentives to limit use of expensive specialty care. Unfortunately, provision of specialty care 
is a very difficult area to monitor, since there is so little agreement as to when referrals to 
specialists are needed. We chose three relatively common procedures in the Medicare 
population-lens replacement, hip and knee replacement, and coronary revascularization--and 
calculated the population-based rate of each procedure. While differences in procedure rates 



Health Economics Research, Inc. Access in Managed Care Plans: 4-4 

crimson\fmal\chap4.wpd\dpb 



Cha P ter4 Discussion 

may in part be attributed to differences in incidence of disease, dramatically high or low 
rates may be cause for concern. Not surprisingly, we found that the surgical rates were 
generally higher in fee-for-service than the HMO divisions. However, this may reflect 
overutilization in fee-for-service, given the incentive structure, as opposed to underutilization 
in managed care. Alternatively, both rates could be appropriate but reflect differences in fc 
casemix. Moreover, given the sample sizes in the HCD and MGD, the number of 
beneficiaries receiving these surgeries in the managed care setting is relatively small and 
unstable from year to year. 

Primary Care 

The proportion of beneficiaries with at least one physician visit during a 12-month 
period is quite high for all three sectors, ranging from 88 percent in fee-for-service to 94 
percent in the HCD. A more striking comparison is found for the percentage of new 
enrollees with at least one visit during the first two months of enrollment. This rate is much 
higher for the HCD than the MGD, and the gap narrows, but does not disappear as the time - 
horizon is expanded. The HCD's high rate reflects its aggressive campaign to triage and 
assess high risk patients. The lower rate for the MGD may reflect or the movement of 
patients into the MGD who join HPHC from another HMO or fee-for-service but do not 
change physicians. These patients would not be assessed as new patients, since they continue 
to visit the same medical group and physician as before joining HPHC. 



Health Economics Research, Inc. Access in Managed Care Plans: 4-5 

crimson\fmal\chap4.wpd\dpb 



Chapter 4 Discussion 

4.2 Implications for Developing a Monitoring System 

This project was intended to serve as a pilot study for determining what measures 
could be constructed-and meaningfully interpreted--with "good" managed care data. It was 
designed to help HCFA in the development of a framework for monitoring managed care. 
This would parallel their ongoing efforts to monitor care in the fee-for-service sector. Hence, 
we conclude with a discussion of "lessons learned" during the course of the study that 
addresses the implications for applying a set of performance measures to other health plans 
or providers. 

4.2.1 Developing the Set of Indicators 

A crucial first step to the study was development of a list of indicators. In addition 
to the indicators included in this report, we also had considered several indicators that were 
eventually deleted from the analysis. Upon further consideration and examination of the 
data, we did not feel that these indicators could be constructed in a meaningful manner. 
However, these indicators are worth documenting so future researchers may be aware of the 
shortcomings we identified. Other indicators on our initial list and our reasons for deleting 
them are as follows: 

Influenza vaccination rate . Constructing the vaccination rate from claims/encounter 
data is problematic since many seniors may receive the flu vaccine at health fairs, senior 
centers, etc. 



Health Economics Research, Inc. Access in Managed Care Plans: 4-6 

crimson\finaI\chap4.wpd\dpb 



Chapter 4 Discussion 

Rates of secondary preventive services for individuals with coronary artery disease 
(CAD) . This indicator contained four components—the percentage of beneficiaries with a 
CAD diagnosis with each of the following during a 12 month period: (1) two or more visits 
with an internist or cardiologist, (2) influenza vaccination, (3) blood pressure screening, and 
(4) serum cholesterol test. The latter three components would be difficult (or for blood v 
pressure screening, impossible) to measure accurately using claims/encounter data. Other 
services for CAD patients such as stress testing and echocardiography are not proven 
appropriate for all patients. While the visit rate could have been constructed, this is very 
similar to other indicators in the analysis, and contributes little to the list. 

Rate of follow-up for abnormal pap smear . The incidence of abnormal pap smears 
is very low among elderly women. While a potentially interesting indicator for the Medicaid 
or commercially insured population, this is a very rare condition among Medicare 
beneficiaries. 

Timing of endarterectomy for individuals with cerebrovascular disease . This 
indicator was defined as the percentage of beneficiaries undergoing carotid endarterectomy . 
who receive surgery within 60 days of the imaging study. It was based on a complicated 
algorithm that was fairly difficult to understand. It also fails to address the important access 
issue of whether individuals receive an endarterectomy when it is clinically appropriate. 
Given the importance of cerebrovascular events among the elderly, we considered as an 
alternative the rate of imaging for patients undergoing a stroke or transcient ischemic attack. 
However, imaging may be inappropriate, depending on the patient's overall condition and 



Health Economics Research, Inc. Access in Managed Care Plans: 4-7 

crimson\final\chap4.wpd\dpb 



Chapter 4 Discussion 

the location of the stroke. Thus, given the lack of detailed clinical data on claims, we 
dropped the indicator entirely. 

Rate of post-myocardial infarction cardiology care . There is no clear evidence as to 
when a specialist (i.e., cardiologist) should be seen by patients with a simple myocardial 
infarction. Primary care physicians may be quite knowledgeable in treating these patients fc 
or (especially in managed care) may consult with a cardiologist without referring the patient 
for a visit. 

Post-menopausal bleeding follow-up . In an earlier era, post-menopausal bleeding 
was uncommon, and if it occurred was considered a clear indication for diagnostic evaluation 
(mainly to look for uterine cancer). Currently, however, many women take hormone 
replacement therapy (HRT), which commonly causes uterine bleeding, sometimes in 
irregular patterns, such that it is a matter of judgment as to when diagnostic evaluation is 
appropriate and when watchful waiting is best. Therefore, variation in rates of evaluation 
for post-menopausal bleeding is likely to be much more sensitive to the frequency with 
which HRT is prescribed and the way in which clinical judgment is exercised in the face of . 
considerable uncertainty and less likely to be an indicator of access to care. 

Fecal occult blood test follow-up . We had originally proposed to determine the rate 
of follow-up for abnormal fecal occult blood tests in the HCD, where test results are recorded 
in the electronic medical record. However, we later learned the tests results are only 
recorded in this format for tests performed in the physician's office and these are generally 
done on symptomatic patients. Results of routine screening tests that are done at home and 



Health Economics Research, Inc. Access in Managed Care Plans: 4-8 

crimson\final\chap4.wpd\dpb 



Chapter 4 _ Discussion 

returned to the lab for analysis are not available in this dataset. Thus, we determined that this 
indicator could only be constructed for a small, non-representative sample of patients. 

Dia gnosis-based surgery rates . Access to elective surgical procedures, such as 
cataract surgery, hip and knee replacement, or coronary revascularization are widely-used 
markers of access to care and have been an important element of international comparisons '- 
of access. We include among our indicators the population-based rates for these procedures 
(Section 3.4. 1 .). We had originally proposed to also construct diagnosis-based rates for each 
of the three groups of procedures. However, upon further consideration, we have decided 
that diagnosis-based rates have several drawbacks. First, if access problems are severe, 
individuals may never have an opportunity to receive medical care to have the diagnosis 
made. This would lead to an overestimate of the proportion of the "eligible" population 
receiving surgery due to undercounting beneficiaries with the condition. Second, the 
propensity to code diagnoses may vary substantially across settings. Chronic conditions such 
as cataract and osteoarthritis progress over time; physicians may differ in whether they code 
the diagnosis the first time it is observed or much later as severity worsens. Third, the coding . 
systems available in our data often do not allow sufficient detail to accurately identify a 
specific diagnosis. For example, the ICD-9 coding allows site of arthritis to be designated 
by use of a fifth digit in the diagnosis code. We had originally thought we could use this 
diagnosis to identify patients with arthritis of the knee or hip who might be candidates for 
joint replacement. However, in the fee-for-service data, we found that 5 percent of claims 
with an arthritis diagnosis have no fifth digit, 34 percent have the fifth digit of "0" 



Health Economics Research, Inc. Access in Managed Care Plans: 4-9 

crimson\final\chap4 wpd\dpb 



Chapter 4 Discussion 

(unspecified site) and 12 percent have the fifth digit of "9" (unspecified multiple sites). 
Given these problems, we felt that the population-based rates were strongly preferable to 
diagnosis-based rates. 

Continuity of care index . We had originally considered constructing the continuity 
of care index for all three sectors. However, constructing this index requires identifying a '«• 
set of primary care visits for each individual, so that the proportion of these visits with each 
provider can be determined. Data from the MGD and fee-for-service sectors both contain 
a specialty code that we had originally thought could be used to identify visits with primary 
care providers. However, both datasets contain a specialty code for "clinic" which is widely 
used and does not provide information on physician specialty. Since it appears that many 
visits with these codes may be for primary care, but we cannot identify exactly how many or 
which ones, we cannot construct the index for continuity of primary care. For the HCD, in 
which specialty is very accurately coded, we did construct the indicator. 

4.2.2 Constructing the Indicators 

Once we had developed the final set of indicators, they were constructed using the 
different claims/encounter databases for Medicare fee-for-service, the HCD, and the MGD. 
In this section, we briefly describe some of the difficulties encountered in developing and 
interpreting the indicators. 



Health Economics Research, Inc. Access in Managed Care Plans: 4-10 

crimson\final\chap4.wpd\dpb 



Chapter 4 Discussion 

4.2.2.1 Reconciling Differences in Coding Systems 

The fee-for-service and MGD data, along with the HCD institutional data, used 
ICD-9 diagnosis and CPT-4 coding. The HCD ambulatory claims used the COSTAR coding 
system that was originally developed by Harvard Community Health Plan. 

Because of the different coding schemes, we were forced to develop comparable 
definitions for identifying diagnoses and procedures for all indicators based on outpatient 
care. In defining the indicators, two questions were considered: 

• Is there an identical (or similar) code in each system? 

• Are physicians equally likely to use the code (given a procedure was performed 
or condition was observed) in each system? 

For many indicators, developing similar definitions was quite straightforward, as 
COSTAR coding corresponded quite closely to ICD-9 or CPT 4 coding. For example, the 
list of codes for colorectal cancer screening tests is fairly extensive, but the definitions of 
codes correspond closely in ICD-9 and COSTAR coding. 

The most difficult definition to develop was for retinal screening for diabetics. The - 
COSTAR system has codes for eye examinations. However, given the payment structure of 
the HCD, optometrists/ophthalmologists have no incentive to code that a specific test was 
performed; rather, they are more likely to code the findings of the test. We found that they 
often coded a diagnosis that would normally require a retinal exam without coding the exam 
itself. Thus, rather than selecting a few COSTAR codes that would correspond to the CPT 
codes for retinal exam, we were forced to rely on a series of diagnostic codes that could only 



Health Economics Research, Inc. Access in Managed Care Plans: 4-11 

crimson\final\chap4.wpd\dpb 



Chapter 4 Discussion 

be found if a retinal exam were performed. If a physician failed to code the exam, and found 
no abnormalities, we may underestimate the numerator for this indicator. 

In fee-for-service, physicians may bill for a visit rather than an eye-exam, since 
payment may differ for the two codes. If this happened, we may also undercount in fee-for- 
service. * 

A second coding issue is the appearance of "rule out" diagnoses in the data. The 
HCD data system allows physicians to mark a diagnosis as being a "rule out"~although it 
is not clear that these are always indicated. The fee-for-service and MGD data have no such 
marker for "rule out" diagnoses, and it is impossible to determine which are intended as 
definitive diagnoses and which are coded as "rule outs." For illnesses which are likely to 
have a high proportion of "rule out" diagnoses in the claims, this difference in coding 
complicates development of similar samples. For the diabetes indicators, we required that 
the diagnosis be attached to a physician claim (rather than, say, a laboratory claim) in an 
attempt to reduce the number of "rule outs." Given the significant number of beneficiaries 
in all three data sets with only one diabetes diagnosis, any attempt to identify all patients with . 
the disease is likely to either miss some true cases or include some rule-out diagnoses. 3 



HEDIS attempts to eliminate "rule-out" diagnoses by requiring that the diagnosis appear twice during the calendar 
year. The disadvantage of this approach is that it may bias estimates of performance indicators upwards, if some 
patients have only one diagnosis because they are low utilizers of care. 

Health Economics Research, Inc. Access in Managed Care Plans: 4-12 

crimson\final\chap4.wpd\dpb 



Chapter 4 Discussion 

4.2.2.2 Variations in Data Set Structure 

In addition to differences in data coding systems, the structures for the data sets 
varied across the three settings. For example, all of the data systems we worked with had 
separate files for inpatient institutional claims. However, the actual claims stored in the 
hospital file differed across the data systems. Initial attempts to locate mammogram codes 
for the MGD identified only 2 percent of women with claims for a mammogram during a 
two-year period, including no claims in 1994. Further investigation revealed that claims for 
Medicare recipients were not located in the ambulatory claims files, but in hospital claims 
files. In contrast, in fee-for-service data, mammography claims can be found in the 
physician/supplier file, the outpatient department file, or both files. 

This example highlights one danger of working with unfamiliar data sets. If all data 
(or virtually all data) are missing, as was the case with mammography in the MGD, it is easy 
to recognize the problem. If some of the data are missing, as was the case in the fee-for- 
service physician/supplier file, it can be much more difficult to recognize that the problem 
exists. 

4.2.2.3 Costs of Processing Data 

The cost of processing claims can be high, especially when it is necessary to search 
through a large database multiple times, for example, to first search an outpatient database 
to identify all claims with a particular diagnosis, and then search again to pull all claims for 
beneficiaries with that diagnosis. 



Health Economics Research, Inc. Access in Managed Care Plans: 4-13 

crimson\ftnaI\chap4.wpd\dpb 



Cha P ter4 Discussion 

For a medical record database, such as HPHC's Automated Medical Record System, 
the cost can be prohibitive, even on relatively small samples of data. Since the data source 
is a medical record, rather than a claim, data processing of relatively small samples of data 
becomes time-consuming and expensive. Thus, in estimating the burden on plans from 
implementing a monitoring system, the data processing requirements should not assume that '* 
all plans have access to claims data and can process data in a similar manner. 

4.2.2.4 Limitations in Sample Sizes 

One of our criteria for selecting indicators was that they be related to a high-incidence 
disease or a high-incidence procedure. Given the limited number of indicators that can be 
monitored, we did not want to select a rare condition (or procedure) upon which to base a 
performance measure. Even using relatively common diseases and procedures, our samples 
were quite small for several indicators in the HCD and the MGD, which had roughly 1 1 ,000 
and 5,500 aged Medicare members, respectively. Sample size decreases even more for 
indicators that require a lengthy continuous enrollment period. Even where overall samples " 
were relatively large, we were often limited in the stratifications that could be made. 

We developed all indicators and presented rates and confidence intervals regardless 
of sample size. (Obviously, the likelihood of detecting statistically meaningful differences 
is much lower for the indicators based on very small samples.) Given the exploratory nature 
of this project, we felt this was an appropriate approach. 



Health Economics Research, Inc. Access in Managed Care Plans: 4-14 

crimson\fmal\chap4.wpd\dpb 



Chapter 4 r*. 

r Discussion 

For a set of performance indicators intended as a "report card," an approach that does 
not rely on audience familiarity with confidence intervals and statistical tests may be more 
appropriate. For example, HEDIS 3.0 specifies that if a measure applies to fewer than 100 
members, the plan should report a 95 percent confidence interval, and that measures based 
on fewer than 100 members should not be used for comparisons among health plans. " 
Moreover, HEDIS specifies that measures should not be reported when there are fewer than 
30 members in the denominator. Our post-depression follow-up measure would not have 
been reported using this criteria, and samples for the myocardial infarction and abnormal 
mammogram follow-ups both fell below the 100 member threshold. 

4.2.2.5 Interpreting the Results 

Claims-based monitoring systems can tell us what occurred in a patient's medical 
care, but not why. For example, the results of our data processing indicated that the rate of 
mammography was much higher in the HCD than in the MGD or fee-for-service. However, 
the claims cannot give us information on whether the difference resulted from provider " 
willingness to encourage mammography, patient willingness to have the procedure, 
availability of convenient locations/hours for mammography services, or some other reason. 
In fact, we believe the difference is largely attributable to the HCD automatic reminder 
system, that prompts physicians when a beneficiary is due to receive a mammogram. 

The advantage of the claims-based system is that it can, at relatively low cost, flag 
areas where the system is doing well or poorly. This allows policy-makers to concentrate 



Health Economics Research, Inc. Access in Managed Care Plans: 4-15 

crimson\final\chap4.wpd\dpb 



Chapter 4 Discussion 

further effort on areas where improvements are needed. By combining a claims-based 
system with other approaches to gauging access and quality, such as surveys and chart audits, 
we can gain a much more complete picture of plan performance. 



4.3 Conclusion 

4.3.1 Generalizability of our Experience 

The purpose of this study was to develop a set of Medicare performance indicators 
that could be applied to managed care plans and to test whether these indicators could be 
implemented using elements available in an HMO data system. This project was intended 
to serve as a pilot study for determining what measures can be constructed, and meaningfully 
interpreted, with "good" managed care data. 

We began the study knowing that our HMO data were of higher quality than that 
found in many managed care organizations. Numerous studies have been published using 
diagnosis and procedure data from the HCD's Automated Medical Record System (studying 
conditions as diverse as streptococcal pharyngitis, hypertension, and bipolar disorder). Data 
from the MGD have not been used for published research to the same extent as data from the 
HCD. However, the plan has used the data bases for its own internal analysis. Thus, 
although we have constructed a set of performance indicators with two types of HMO data, 
it is not clear whether the data systems of other managed care organizations can support the 
same types of analysis. Many pressures (including HEDIS) are pushing managed care 



Health Economics Research, Inc. Access in Managed Care Plans: 4-16 

crimson\final\chap4.wpd\dpb 



Chapter 4 Discussion 

organizations to improve their data systems. Thus, construction of performance indicators 
is much more feasible than it would have been even a few years ago 

4.3.2 Next Steps 

For this project, we developed a set of 19 performance indicators, several of which 
were constructed using alternate methodologies (for example, varying the episode length). 
While we constructed multiple rates in order to test the sensitivity of our results to varying 
definitions, it would be desirable to determine the preferred definition that would be reported 
as part of the performance monitoring system. 

More importantly, it would be desirable to replicate this study using data from other 
health plans. Using data from two divisions of HPHC, we have found that our indicators can 
be constructed, and comparisons among the two divisions and fee-for-service practice show 
meaningful differences in the performance of the three sectors. We have also found, 
however, that differences in databases can complicate construction and interpretation of the 
indicators. Extending the work to include data from other health plans would be the next " 
step towards developing these indicators into a monitoring system for managed care 
performance. 



Health Economics Research, Inc. Access in Managed Care Plans: 4-17 

crimson\final\chap4.wpd\dpb 



REFERENCES 



Berenson R and J Holahan: Using a New Type-of-Service Classification System to 
Examine the Growth in Medicine Physician Expenditures, 1985-1988. 
HCFA Cooperative Agreement No. 17-C-99473/3, The Urban Institute, 
Washington, DC, December 1990. 

Bernstein AB, GB Thompson, and LC Harlan: "Differences in Rates of Cancer 

Screening by Usual Source of Medical Care." Medical Care 29(3): 196-209, 
March 1991. 

Billings, J. et ah "Impact of Socioeconomic Status on Hospital Use in New York City." 
Health Affairs 12:162-173, 1993. 

Brown RS, et ah Do Health Maintenance Organizations Work for Medicare? Health 
Care Financing Review 15 (1): July 23, 1993. 

Dayhoff DA and J Cromwell: Implementing Findings on Volume/Quality. Final Report, 
HCFA Cooperative Agreement No. 99-C-98526/1-08, October 1994. 

Federal Register 59(169). Department of Health and Human Services: Washington, DC. 
September 1, 1994. 

1991 Healthy People 2000, Department of Health and Human Services. Washington, 
DC: U.S. Government Printing Office. 

Docteur ER, DC Colby, and M Gold: "Shifting the Paradigm: Monitoring Access in 
Medicare Managed Care." Health Care Financing Review 17(4):5-20, 1996. 

"HCFA Should Release Data to Aid Consumers, Prompt Better HMO Performance." 
GAO/HEHS-97-23. General Accounting Office, Washington, DC, 1996. 

Goldzweig, CL, et ah. Variations in Cataract Extraction Rates in Medicare Prepaid and 
Fee-for-Service Settings. JAMA 277:1765-1768, 1997. 

Greenfield S, et ah: "Variations in Resource Utilization Among Medical Specialties and 
Systems of Care: Results from the Medical Outcomes Study." JAMA 267:1624- 
1630, 1992. 

Kannel WB and TJ Thorn: "Incidence, Prevalence, and Mortality of Cardiovascular 
Diseases." In Hurst (editor). The Heart . 8th Edition. McGraw Hill, 1994. 

Manning WG, et ah: "A Controlled Trial of the Effect of a Prepaid Group Practice on 
Use of Services." New England Journal of Medicine 310:1505-1510, 1984. 



Health Economics Research, Inc. 

cnmson\fmal\ref.wpd\dpb 



Access to Managed Care Plans: R-l 



References 



Miller RH and HS Luft: "Managed Care Plan Performance Since 1980: A Literature 
Analysis." JAMA 271:1512-1519, 1994. 

Mitchell JB: "Impact of the Medicare Fee Schedule on Access to Physician Services." 
Annual Report to Congress. Monitoring the Impact of Medicare Physician 
Payment Reform on Utilization and Access . Department of Health and Human 
Services. Washington, DC: 1994. 

Morgan RO, et al.: "The Medicare-HMO Revolving Door: The Healthy Go in and the 
Sick Go Out." New England Journal of Medicine 337(3):169-175, 1997. 

Monitorin g Access of Medicare Beneficiaries . Physician Payment Review Commission: 
Washington, D.C. 1995. 

National Committee for Quality Assurance, Health Plan Employer Data and Information 
Set, (HEDIS 3.0). 1996. 

Nelson L, et al: Access to Care in Medicare HMOs, 1996. Health Affairs 16 (2): 148- 
156, 1997. 

"Public Sector Contracting Report." National Health Information L.L.C. 3(5):65-80, 
1997. 

Riley GF, MJ Ingber, and CG Tudor: "Disenrollment of Medicare Beneficiaries from 
HMOs." Health Affairs 16(5):1 17-124, 1997. 

Riley GF, et al: "Stage of Cancer at Diagnosis for Medicare HMO and Fee-for-Service 
Enrollees." American Journal of Public Health 84(10):1598-1604, 1994. 

Rosenbach ML and R Khandker: "Changes in Utilization, Access, and Satisfaction with 
Care among Noninstitutionalized Medicare Beneficiaries." Annual Report to 
Congress. Monitoring the Impact of Medicare Physician Payment Reform on 
Utilization and Access . Department of Health and Human Services. Washington, 
DC: 1994. 

Rowland D and B Lyons: "Mandatory HMO Care for Milwaukee's Poor." Health 
Affairs 6(1):87-100, 1987. 

Siu AL, et al.: "Choosing Quality of Care Measures Based on the Expected Impact of 
Improved Care on Health." Health Services Research 27(5):6 19-650, 1992. 



Health Economics Research, Inc. 

crirason\final\ref.wpd\dpb 



Access to Managed Care Plans: R-2 



References 



Spitz B: "When a Solution is Not a Solution: Medicaid and Health Maintenance 

Organizations." Journal of Health Politics Policy and Law 3:497-518, 1979. 

Ware JE, et al: "Differences in 4-year Health Outcomes for Elderly and Poor, 

Chronically 111 Patients Treated in HMO and Fee-for-Service Systems." JAMA 
276(13):1039-1047, 1996. 

WennbergJE: The Dartmo uth Atlas of Health Care in the United States . Center for 
Evaluative Clinical Sciences. Dartmouth Medical School, 1996. 



Health Economics Research, Inc. 

crimson\finaI\refwpd\dpb 



Access to Managed Care Plans: R-3 



•■ 



Appendix A 



Appendix A 

Profile of Harvard Pilgrim Health Care 
The managed care data for this study come from Harvard Pilgrim Health Care 
(HPHC), formerly know as Harvard Community Health Plan. This appendix provides a 
profile of the plan, including its provider structure, membership, benefits, capacity and 
service delivery, medical management systems, care for new members, access, and member fc 
satisfaction measurement. The purpose of the case study is three-fold. First, it helps us to 
better understand the different structures and incentives in the two divisions that provide data 
for the study. Second, it provides qualitative information on two components of our 
conceptual framework that we are not measuring empirically (resource availability and 
satisfaction). Third, it helps in understanding mechanisms used by the plan to 
monitor/promote access which may help explain plan performance. The appendix is 
structured in a question and answer format. 

1.0 Overview 

What is the corporate structure of HPHC? How did it evolve? 

In 1966, the Dean and his colleagues at Harvard Medical School began planning for 
New England's first prepaid group practice, the first in the nation to be affiliated with an 
academic medical center. Harvard Community Health Plan (HCHP), the result of this effort, 
opened its doors at one location in 1969. During the ensuing years more sites were 
established and membership grew. HCHP merged with Multigroup Health Plan, a group 
model HMO serving the suburban Boston area, in 1986 to become a mixed-model 
staff/group HMO. In 1990, the Rhode Island Group Health Association, a predominantly 



Health Economics Research, Inc. Access in Managed Care Plans: A-l 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

staff-model HMO operating throughout Rhode Island and southeastern Massachusetts, 
affiliated with HCHP and became the New England Division. In 1995, HCHP affiliated with 
Pilgrim Health Care, a large managed care organization in the region, to form Harvard 
Pilgrim Health Care. 

HCHP (as we will continue to refer to the organization prior to the merger with 
Pilgrim Heath Care) included three divisions. The Health Centers Division, the original staff 
model HMO, has grown to 14 sites in the greater Boston area (see Exhibit A-l). The New 
England Division, as described above, was formerly an independent HMO in Rhode Island. 
The Medical Groups Division, in which HCHP contracts with group practices, is the fastest- 
growing division as it continues to expand its affiliation with existing practices throughout 
New England. The HCHP enrollment area for the 1994-95 period included portions of New 
Hampshire, Massachusetts, and Rhode Island. Exhibit A-2 shows the clinical sites that 
treated HCHP's Medicare enrollees during this period. 

How are physicians in the HCD and MGD paid? 

The Health Centers Division (HCD) is a staff model HMO. Nearly all care is by 
salaried staff of HPHC, including that by primary care and most specialist physicians, such 
as surgeons performing cataract surgery or hip replacement. HCD does not require a full- 
time physician on staff to care for less common conditions, for example, to perform cardiac 
surgery. Instead, the HCD contracts with local physicians for part of their time. For very 
uncommon needs, the HCD purchases services on a fee-for-service basis from highly 
specialized physicians. 



Health Economics Research, Inc. Access in Managed Care Plans: A-2 

crimson\final\apndx-a.wpd\dpb 



Appendix A 



Profile of Harvard Pilgrim Health Care 



Exhibit A-1 HCHP Health Center Locations 



New Hampshire 



Rhode 
Island 



(■!■) Health Center Locations 




Health Economics Research, Inc. 

crimson\final\apndx-a.wpd\dpb 



Access in Managed Care Plans: A-3 



Appendix A 



Profile of Harvard Pilgrim Health Care 



(0 

c 

O 

*-» 

o 
o 





* WW 

> 

o 



c 
w 

Q. 



o 

■ ww 



X 

o 



CM 

I 

< 



X 
LU 




Health Economics Research, Inc. 

crimson\final\apndx-a.wpd\dpb 



Access in Managed Care Plans: A-4 



Appendix A 



Profile of Harvard Pilgrim Health Care 



In the Medical Groups Division (MGD), HPHC contracts with groups of physicians 
who are geographically dispersed throughout the region and are not on HPHC's staff. 
Physicians are paid through a variety of negotiated arrangements. For large multispecialty 
groups, capitation payments are negotiated for ambulatory and professional care and bonuses 
are paid for clinical and administrative performance, such as member satisfaction, meeting 
appropriateness of care criteria, compliance with the drug formulary, and prior notification 
of elective hospital admissions. Small multispecialty groups have arrangements similar to 
large multispecialty groups except capitation rates are age- and sex-based and not negotiated. 
For single specialty groups, primary care is capitated on an age- and sex-adjusted formula. 
Specialty care is paid for from a risk pool, there is a separate budget for hospital care, and 
bonuses are paid as for large multispecialty groups. For all groups there is a ceiling on the 
losses that can be incurred by an individual group. 



2.0 Membership and Enrollment 

What does membership in HPHC look like? 

The enrollments during the study period in the Health Centers and Medical Groups 
Divisions were as follows: 



Division 



Total Enrollment 
fas of 9/951 



Medicare 
(as of 5/951 



Health Centers 



302,056 



11,047 



Medical Groups 



186,027 



5,493 



Health Economics Research, Inc. 

crimson\final\apndx-a.wpd\dpb 



Access in Managed Care Plans: A-5 



Appendix A Profile of Harvard Pilgrim Health Care 

Just over half of the total enrollment was female, and 5 percent was over age 65. 
Enrollees were multi-ethnic; 74 percent were White, 16 percent African American, 4 percent 
Hispanic, 1 percent Asian, and 6 percent other/unknown. Membership included many 
immigrants from Europe and Latin America, reflecting the ethnic make-up of the Boston area 
as a whole. 

L 

Table A-l presents gender and age breakdowns for Medicare enrollees in the Health 
Centers and Medical Groups Divisions, and for all Medicare beneficiaries nationally. HCHP 
membership contained a slightly lower proportion of female and more enrollees in the 65-69 
age group than the overall Medicare population. The plan contained a lower proportion of 
older enrollees, particularly among the oldest age groups. While 24 percent of Medicare 
beneficiaries are age 80 or older, only 12 percent of HCD and 16 percent of MGD Medicare 
enrollees were in this age group. 

What type of information is available to new members? 

New members are sent a packet of information including HPHC's philosophy of 
health care, benefits, monthly costs, options (such as drug or dental coverage), and a 
physician directory (with locations). 

How are new members assigned to primary care providers? 

In the Health Centers Division, new members receive a listing of all physicians with 
a brief description about each physician's background. Members service representatives at 
each of the 14 centers take telephone calls from new members and direct them to clinicians 



Health Economics Research, Inc. Access in Managed Care Plans: A-6 

crimson\finaI\apndx-a.wpd\dpb 



Appendix A 



Profile of Harvard Pilgrim Health Care 



Table A-l 



Demographic Characteristics of Aged Medicare Enrollees: 
Nationally and for HPHC Enrollees 



United 
States 



FFS 
Area 



Harvard Pilgrim Health Care 



Health Centers 
Division 



Medical Groups 
Division 



Percent Female 


60% 


62% 


56 


Percent bv Age Category: 








65-69 


30 


23 


42 


70-74 


26 


29 


28 


75-79 


20 


23 


18 


80-84 


13 


15 


8 


85+ 


11 


10 


4 



55 % 



44 
24 

16 
9 

7 



NOTES: 1. National values are for all aged Medicare beneficiaries. 

2. FFS area represents the portion of New England included as our comparison group. 

SOURCES: 1994 Data Compendium of the Health Care Financing Administration; Harvard Pilgrim Health Plan enrollment data. 



Health Economics Research, Inc. 

crimson\final\apndx-a.wpd\dpb 



Access in Managed Care Plans: A-7 



Appendix A Profile of Harvard Pilgrim Health Care 

with open practices, taking into consideration members' preferences (such as for gender or 
language spoken). 

The various provider groups in the Medical Groups Division do not have any shared 
way of assigning new patients or monitoring availability and caseloads. Each group arranges 
for the assignment of new members in its own way. 

What types of risk assessment are performed on new Medicare members? 

Members of a geriatrics assessment team survey all new Medicare members by 
telephone as part of a risk assessment protocol. The instrument includes questions about 
self-rated health, treatment of illness, number of medications, recent hospitalizations, 
activities of daily living and other risk factors. A specially trained nurse reviews all 
completed risk-assessment questionnaires, guided by definitions for risk strata. Standards 
for scheduling an initial visit are: very high risk, within 10 working days; high risk, 25 
working days; moderate risk, 1-3 months; and low risk, 2-6 months. The risk assessment 
questionnaire does not take into account non-medical factors (e.g., need for transportation, 
existence of family supports) that might affect the need for health services, but specially 
trained nurses can over-ride the score in assigning risk. Risk status information is shared 
with the member's primary physician, a First Seniority Committee at each HCD site, and in 
some cases, case managers at each site. If the risk assessment review suggests that the 
member is at high risk the risk assessment team may arrange for some of their care before 
the initial visit with a clinician. 



Health Economics Research, Inc. Access in Managed Care Plans: A-8 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

3.0 Benefits 

What benefits were available for Medicare beneficiaries during the J 994-95 study period? 

HPHC's main insurance plan for Medicare patients is First Seniority, HPHC's 
Medicare "risk" contract. This is the plan actively being marketed today. During our 1994- 
95 study period HCHP offered three other Medicare products. CarePlus was available only 
to members of the New England Division in Rhode Island and southeastern Massachusetts. 
The two remaining plans, Plan65 and Senior Care, were both phased out beginning in 1995 
with enrollees being converted to First Seniority. Exhibit A-3 indicates the clinical sites 
available to members of each plan as of 1994. 

Enrollees in the HCHP plans and those in Medicare fee for service (FFS) plans 
received different benefits (although benefits in the three HCHP plans were quite similar). 
These are detailed in Tables A-2 through A-6 at the end of this appendix. For outpatient 
physician care, HCHP members were responsible for only a $5 copay, while FFS Medicare 
enrollees had a $100 dollar deductible and pay 20 percent of allowed charges thereafter for 
all outpatient care. HCHP also provided full coverage (no copay) for laboratory and x-ray 
services, durable medical equipment, ambulance service, for which FFS enrollees were also 
subject to the Part B deductible and copay s. HCHP covered numerous preventive and 
screening services such as routine physical exams, hearing exams, eye exams, and 
immunizations that were not covered by FFS. In addition, HCHP offers optional prescription 
drug coverage which was available to FFS enrollees only through a supplemental plan. 



Health Economics Research, Inc. Access in Managed Care Plans: A-9 

crimson\final\apndx-a.wpd\dpb 



s.sc 

g » 
9 JS 

sr 2 
•o a 
3 o 

o. ** 

B 

-1 

O 



Exhibit A-3 HCHP Medicare Plan Provider Locations 



a 
a 



2 

» 

a 
to 

Q. 



n 

2 
ST 



I 

o 




Provider Locations 

Care Plus Locations 
First Seniority Locations 
Plan 65 Locations 
SeniorCare Locations 



-i 

o 

B 



s 



n 



Appendix A Profile of Harvard Pilgrim Health Care 

Physician inpatient services and hospital care were covered in full for HCHP 
enrollees, as were home health services and skilled nursing facility stays (with some 
limitations on covered days per benefit period). For outpatient mental health and substance 
abuse services, HPHC members paid $5 per visit for the first 8 visits, $35 per visit for the 
9th through 20th individual session, and 50% of all charges for all visits thereafter. FFS 
enrollees were subject to the Part B deductible (if they have not already paid it) and pay 50% 
of charges thereafter. Inpatient mental health coverage was similar in HPHC and FFS. 

4.0 Capacity and Service Delivery 

How does HPHC monitor a provider 's caseload size? Are there caseload expectations or 
limits for physicians? 

In the Health Centers Division, caseload is tracked by computer. If a physician's 
caseload is low, the physician is not allowed to close the panel. The current panel target for 
commercial members is 1,600, reduced from 1,800, to allow physicians more time for case 
management. Each full-time physician is counted as one FTE, while each nurse practitioner 
or physician assistant is counted as one half an FTE for panel size calculations. HPHC is - 
developing a more complex metric for establishing target panel sizes, taking into account 
members' age, gender, visit rates and possibly certain conditions. Panel targets are smaller 
in proportion to the number of Medicare members; if the panel were entirely Medicare, the 
panel size would be 600-700. 

HPHC is not involved in caseload monitoring for the Medical Groups Division. The 
groups make their own decision about whether to keep their panels open or to close them. 
Unlike some HMOs, HPHC does not pressure groups to be open to new members at all 



Health Economics Research, Inc. Access in Managed Care Plans: A-ll 

crimson\final\apndx-a. wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

times. As a result, busy groups will sometimes close to HPHC or be effectively closed, i.e., 
accepting new patients but with very long waits for appointments. 

How is productivity monitored? Are there productivity expectations for physicians? 

The standard for the Health Centers Division was historically 30-32 bookable hours 
per week, equaling roughly 80 visits per week. These standards are changing with the 
introduction of "designated rounders," also called intensivists, who are assigned to follow 
hospitalized members during their inpatient stays. For example, there are typically 30-40 
HPHC members at the Brigham and Women's Hospital each day who would be followed by 
"rounders" rather than by their regular primary care physicians. Thus, a primary care 
physician's bookable hours would increase as they are relieved of hospital duty. 

In the Medical Groups Division, there are no productivity requirements set by HPHC. 
Groups develop their own internal arrangements to monitor their individual physicians. 

How does HPHC credential new physicians or practices? 

In the Health Centers Division the application process for physicians is handled at the 
organizational level. The applicant must submit proof related to education, residency, 
license, specialty, national medical boards, and any credentials at other institutions. Board 
eligibility is required and new hires are expected to become board certified within a specified 
time frame. Previous hires without board certification are grandfathered. HPHC checks the 
National Practitioner Data Bank, the Board of Registration, and references from employers 
or colleagues. The actual hiring decisions are made by the individual health centers, with 
organizational approval. Ongoing staff need to be recredentialled every two years. 

Health Economics Research, Inc. Access in Managed Care Plans: A-12 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

Physicians are reviewed annually, with respect to clinical quality, service to patients, service 
to the team, and cost effectiveness/resource utilization. 

HPHC has an extensive process for credentialing new groups wishing to join the 
Medical Groups Division. Standards for a group to join HPHC include the following: 

• the group must demonstrate willingness to participate in managed care; 

• the group must have hospital admitting privileges; 

• the group must provide self-contained 24-hour coverage, (between-group 
coverage can only be provided if both groups belong to HPHC); 

• the group must provide the full range of services consistent with primary 
care and the practitioner's specialties; 

• the group must have a network including a full range of specialists. The 
group can either have their own referral network or be willing to accept 
the specialists that HPHC assigns to them. 



Interested groups submit a profile of the practice and the individual physicians to 
HPHC. This profile includes the nature of the practice, hours of operation, coverage, 
education, experience, and board certification status of physicians. Physicians must be 
board eligible, but are not required to be board certified. If an individual member of a 
practice is unacceptable, HPHC will not contract with the group. 

After reviewing the profile, HPHC staff make a series of on-site visits to the group. 
The HPHC staff includes business personnel, the medical director responsible for the group, 
and support personnel. HPHC staff look at the site, review the records and record 



Health Economics Research, Inc. Access in Managed Care Plans: A-13 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

management practices of the group, and provide information to the group, about HPHC 
practices. 

How is provider availability for new patients assured? 

When practices are closed to new patients, members' choices are constrained. HPHC 
has taken steps to assure that as many practices as possible are open. With the advent of First 
Seniority, the Plan undertook a site by site analysis of physicians' availability for new 
patients. It found several centers with a high proportion of closed practices (in July 1995 the 
range was - 60%) and recommended that they review their methods for managing panels. 
At monthly intervals, an administrator for each department at each site distributes charts that 
show each clinician's availability--e.g., waiting time to next available short (return) 
appointment and to next available long (initial) appointment. Actual availability for 
physicians and nurse practitioners in each specialty is displayed along with standards for 
each. Some centers have also prepared summaries of each physician's availability over 
several consecutive months. 

5.0 Medical Management Systems 

What type of reminder system does the Plan have to ensure services are provided? 

The Health Centers Division's electronic medical record displays a reminder system 
at the beginning of the medical record at each encounter. The conditions included in the 
reminder system are determined system-wide by each specialty. Three kinds of automated 
reminders exist: for preventive health services such as mammography and Pap smears; 



Health Economics Research, Inc. Access in Managed Care Plans: A-14 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

periodic reminders such as for influenza vaccine every fall; and registry-based reminders 
such as for diabetic eye examinations. HPHC is considering developing reminder systems 
for patients who have missed scheduled visits. 

Practices in the Medical Groups Division do not have an electronic appointment and 
medical record system in common and as a result, do not have such well developed reminder 

L 

systems (although a variety of systems exist at the various sites). HPHC issues batch lists 
of members who have HEDIS "defects" which serve as reminders to providers that services 
are required. 

What type of case managers does the Plan have? 

Most Health Center Division sites have case managers who become involved with 
members identified as high risk or at the request of their physicians. The target staffing ratio 
is one case manager per 1,500 Medicare members (compared with 1:2,000 for other 
members). An automated case management information system for extended care facilities 
is being developed. 

In MGD, HPHC keeps 20% of the Medicare capitation for administrative expenses. 
Eighty percent of this share, 0. 1 6% (.8 x .2), is spent on case management. There are 30 RN 
case managers, each of whom are assigned to particular medical groups. Since there is very 
little turnover in this department, the case managers are highly experienced in case 
management for the commercial population. They are developing their expertise in case 
management for Medicare members. Many worked in home health agencies before coming 
to HPHC. While each medical group is assigned a case manager, the case managers are not 



Health Economics Research, Inc. Access in Managed Care Plans: A-15 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

onsite and spend much of their time case finding and preparing discharge plans at local 
hospitals. Primary care physicians also refer to case managers directly. 

What clinical management protocols have been developed for chronic illness? 

The Health Centers Division has developed a set of guidelines for preventive care and 
the care of many acute and chronic conditions. The guidelines are prepared by the Clinical 
Quality Management group in collaboration with clinicians with specific expertise or a strong 
stake in the guideline. Guidelines are made available through a hard copy, loose-leaf book 
and sent to all clinicians. Guidelines are updated periodically. A few of the guidelines are 
specifically for the care of elderly patients. 

Because 30-40 of the beds at Brigham and Women's Hospital are occupied by Health 
Center Division members on any given day, the hospital and HPHC have collaborated in the 
design of "critical pathways" for specific diagnoses. Developed by multidisciplinary teams, 
the pathways describe the usual time course for procedures, medications, and transfers to 
simpler facilities for average patients with specific conditions such as acute myocardial 
infarction. The protocols also include elements of follow-up care. 

The Medical Groups Division offers incentives to the groups to use chronic disease 
management modules developed by HPHC. In addition, groups develop their own 
guidelines. They track health and financial outcomes of care and use them to select diseases 
for which to develop guidelines. 



Health Economics Research, Inc. Access in Managed Care Plans: A-16 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

What is HPHC's philosophy regarding geriatric training and use of geriatricians? 

HPHG believes that most care of the elderly should be by primary care physicians 
(general internists) and not geriatricians. To transfer members to geriatricians when they 
reach age 65 years would result in discontinuity of care and require a massive reworking of 
the workforce as more elderly patients become members. The role of geriatricians is to raise 
the level of understanding of geriatric care among generalists. The system of care is thus 
multi-tiered: most elderly members are seen by general internists; some of these patients are 
seen in consultation with geriatricians; and a very few are directly under the care of 
geriatricians. To implement this strategy, HPHC seeks to have a geriatrician (either by 
specialty training or by retraining) at each of the large sites, to act as a consultant and to care 
for the most complex geriatric patients. It has also redeployed some general internists to 
extended care facilities, where they specialize in post acute care. 

To increase geriatric competency among HPHC clinicians as a whole (both Health 
Centers Division and Medical Groups Division), the Plan has begun a special geriatrics 
education program known as the«Medicare Education Partnership Program. The program 
organizes large conferences and small teaching sessions at the various sites, offered free of 
charge to all clinicians in HPHC. This initiative was motivated by HPHC's earlier 
experience with a small number of capitated Medicare patients. Primary physicians had been 
overwhelmed by the needs of Medicare patients, and the program had not been financially 
successful because HPHC has no special plan for managing their care. 

Assuring geriatrics competency in the Medical Groups Division is complex because 
physicians are affiliated with HMOs other than HPHC and may receive geriatrics education 



Health Economics Research, Inc. Access in Managed Care Plans: A-17 

crimson\final\apndx-a.wpd\dpb 



Appendix A 



Profile of Harvard Pilgrim Health Care 



from the other HMOs or other courses. Therefore, it is difficult to track what MGD 
physicians have had in the way of geriatrics education. 

6.0 Mechanisms to Monitor and Promote Access 

Does HPHC have standards regarding waiting times? How are waiting times monitored? 
Standards for waiting times were developed in 1996 (after the study period for this 
project). The new standards for appointment access are as follows: 



Primary Care 




Routine non-symptomatic appointments 
(e.g., check-ups, immunizations) 


<= 30 calendar days 


Non-urgent symptomatic (e.g., non-acute 
symptoms) 


<= 7 calendar days 


Urgent (e.g., acute symptoms) 


<= 24 hours 


Emergency 


Immediate 



Specialty Care 




Initial non-urgent appointment (e.g., non- 
acute symptoms) 


<= 14 calendar days 


Initial urgent appointment (e.g. acute 
symptoms) 


<= 7 calendar days 



Health Economics Research, Inc. 

crimson\finaI\apndx-a.wpd\dpb 



Access in Managed Care Plans: A-18 



Appendix A 



Profile of Harvard Pilgrim Health Care 



Mental Health Care 




Routine, non-symptomatic, preventive care 
(e.g., non-acute counseling for vocational 
issues) 


<= 7 days 


Initial non-urgent symptomatic care (e.g., 
chronic but non-acute symptoms or poor 
functioning) 


<= 7 days 


Follow-up non-urgent symptomatic care 


<= 14 days 


Initial and follow-up urgent appointments 
(e.g., patients at risk of serious deterioration 
of functioning or in acute crisis) 


<= 24 hours 


Emergency (i.e., risk of imminent physical 
harm to self or others or psychosis) 


Immediate 



The computerized appointment system can monitor waiting times in HCD. 

What provisions does HPHC make for after-hours and emergency care? 

Arrangements for after hours care differ in the HCD and the MGD. The Health 
Centers Division has selected three centers (Kenmore, Somerville, and Wellesley) that are 
geographically dispersed to provide extended hours. These centers are open late in the 
evenings and during the day on weekends. Members are encouraged to call after-hours either 
to make an appointment for urgent care or to have a nurse call them back. Some patients 
show up at the urgent care clinics without calling, which is also allowed. Regular staff and 
contract staff (such as some nurses who only work Saturdays and Sundays) provide after- 
hours care. The patient's primary care physician routinely receives a report of the encounter 



Health Economics Research, Inc. 

crimson\final\apndx-a.wpd\dpb 



Access in Managed Care Plans: A-19 



Appendix A Profile of Harvard Pilgrim Health Care 

within two days. However, if follow-up is necessary, urgent care will call the primary care 
office. 

To assist members who need after-hours care, HPHC staffs a liaison nurse or 
physician at Brigham and Women's Hospital. The organization has not had a problem with 
members using the emergency room inappropriately. The types of cases seen in the ER vary. 
Before midnight, patients have the option to go to urgent care, and normally would only use 
the ER in case of a true emergency. Thus, someone with relatively minor injuries from a 
household accident would be referred to urgent care in the evening, while someone with a 
suspected heart attack would be told to go to the ER. After midnight, anyone whose 
condition is serious enough that they cannot wait until morning for care would be referred 
to the ER. If necessary, physicians can authorize an ambulance to bring the patient to the 
hospital. 

Within the plan's coverage area, the hospital providers used for elective care are 
restricted by the plan, but any hospital can be used in an emergency. Thus, for a true 
emergency, patients would be sent to the closest appropriate hospital. However, if they are 
admitted to a hospital not affiliated with HPHC, their physicians may arrange a transfer to 
a hospital with which HPHC contracts. 

Within the MGD, all groups are required to arrange 24-hour coverage for HPHC 
members. Only physicians affiliated with groups having an HPHC contract are allowed to 
provide coverage, to ensure that all physicians potentially treating members meet HPHC 
standards. Beyond these types of basic restrictions, arrangements for after hours care differ 
among groups, as they are allowed to determine how coverage will be provided. 



Health Economics Research, Inc. Access in Managed Care Plans: A-20 

crimson\fmal\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

Urgent care outside the HPHC enrollment region is covered, with "urgent" defined 
as care that is not "preventive, foreseen or routine." Some concerns have been raised 
regarding exactly what out of area care should be covered for patients with chronic 
conditions. This issue is particularly relevant among the elderly who have a high prevalence 
of chronic conditions and are more likely than younger beneficiaries to be out of area on 
extended trips. For example, if a beneficiary with a chronic condition were in Florida for 3 
months during the winter, what care would be "foreseen" versus "unforeseen?" While this 
has not been a major issue for the plan, it is recognized as a gray area that could become 
increasingly common with the growth in the number of Medicare enrollees. 

How do members get referrals to specialists? 

HPHC enrollees do not need a primary care referral for dermatology, mental health 
or obstetrics/gynecology care. For other specialty referrals within the HCD network, a 
primary care physician has to authorize a referral for the first visit to a given specialist; 
thereafter, the specialist freely determines ongoing care. For out of network referrals, the 
primary care physician refers for a limited number of visits (which can be extended if 
necessary). 

In the MGD, the primary care groups subcontract to specialists on a contracted fee- 
for-service basis. With few exceptions, the HCD specialists are not available to MGD 
(exceptions include second opinions, and HCD's oncology service.) 

The specialist appointment is either made by the medical assistant, through interoffice 
mail or if emergent, physician to physician via phone call or page. If a member does not 



Health Economics Research, Inc. Access in Managed Care Plans: A-21 

crimson\finaI\apndx-a wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

follow through with a specialist within 3 months of the referral, the primary care physician 
is notified. 

How do members gain access to mental health services? What mental health services are 
available? 

Members do not need a referral to schedule an initial appointment for mental health v 
care. However, the mental health benefit is tightly managed and oriented toward brief 
interventions for individual therapy and groups for the chronically mentally ill. Mental 
health care is provided by professionals with various training including psychiatrists, 
psychologists, masters in social work and advance practice nurses. 

The coverage for outpatient mental health and substance abuse services differs from 
other outpatient services. The coverage for each calendar year is: $5 copay per visit for 
visits 1-8; for visits 9-20, copay of $15 per visit for a group session or $35 per visit for an 
individual session; after the 20th visit, copay of 50 percent of the full charge per visit. 

What types of enabling services (translation, transportation) are available to Medicare . 
members? 

The HPHC Office of Diversity helps coordinate translation services. Most centers 
have staff who speak a variety of languages (this information is kept by each center), and 
HPHC currently is pilot testing the use of on-call translators in a few centers with the highest 
proportion of non-English speakers. Otherwise, translation services are provided by AT&T. 

HPHC does not provide any special transportation services for Medicare members, 

but case managers assist in coordinating services (such as the Senior Shuttle) that are 

available locally. Additionally, some geriatricians make home visits. 

Health Economics Research, Inc. Access in Managed Care Plans: A-22 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

7.0 Measurement of Member Satisfaction 

What approaches does HPHC use to measure satisfaction among members? 

HPHC has been surveying patients' satisfaction for many years, including a member 
survey and surveys tied to specific visits. In the past, the visit survey was handed out to 
members during office visits to primary care physicians and specialists. The rationale for 
sampling visits rather than members was to anchor responses to a specific encounter rather 
than to elicit member satisfaction in general. The 21 question instrument asked about 
satisfaction with the length of time to get an appointment, time of day, waiting during the 
visit, behavior of clinicians and support staff, and overall satisfaction. In 1994, 29,000 visit 
surveys were completed. Most respondents reported being satisfied, however there is 
potential selection bias in who chose to respond and the wording of the questions may also 
influence the responses. 

HPHC is testing use of a mail survey to a random sample of members with recent 
visits. This will allow HPHC to calculate response rates and also to avoid the potential 
selection bias associated with the previous approach. Additionally, a mail survey is more 
practical given over 4000 provider sites. 

HPHC surveys about 1 50 - 200 visits per physician annually. Pooled together within 
a department, this is enough to produce estimates of satisfaction that are statistically 
meaningful on the department level. Information on individual physicians, while statistically 
unstable, provides the individual physicians some feedback. In the past, satisfaction 
information was only sent to department managers and chiefs. Now, the information about 
individual physicians and comparative information by department and across sites is sent to 



Health Economics Research, Inc. Access in Managed Care Plans: A-23 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

individual physicians as well. HPHC is also working with other HMOs, through NCQA, to 
develop a common satisfaction survey instrument. 

HPHC's member survey is more comprehensive than the visit survey and is 
administered through the mail. The member survey collects more general impressions about 
the delivery system in the areas of medical care, access, support staff, coverage, cost and 
administration. Comparisons of survey results are made across years. The response rate is 
approximately 35 percent. 

Another approach to monitoring member satisfaction is to review members' reasons 
for voluntary disenrollment. All members who disenroll must provide documentation of 
their decision in writing and HPHC asks for their reason for disenrollment in a specific form 
for this purpose. Although not all disenrollees respond to this part of the form, some patients 
who do not fill out the form volunteer their reasons. All disenrollment information is 
reviewed in a single office and reasons for disenrollment classified into crude categories. 
Few voluntary disenrollments are because of dissatisfaction. Those disenrollees who do cite 
dissatisfaction as their reason for leaving HCHP are primarily concerned with either access 
to specialists or with administrative issues regarding coverage of emergency room use. More 
detailed, system- wide reports of disenrollment are being developed. 

What types of surveys does the plan use to measure satisfaction among Medicare 
beneficiaries? 

There is no separate survey instrument for Medicare members. However, information 

on age is included so it is possible to examine the Medicare subgroup or elderly cohort 

separately. 



Health Economics Research, Inc. Access in Managed Care Plans: A-24 

crimson\final\apndx-a.wpd\dpb 



Appendix A Profile of Harvard Pilgrim Health Care 

What actions, if any, have been taken to improve member satisfaction? 

In HCD, satisfaction information is fed back to individual physicians who discuss it 
with their department chiefs to identify areas for improvement. Summary information is also 
sent to medical directors and corporate management, who use the data to identify patterns 
of dissatisfaction and to plan remedial action. HPHC plans to use patient satisfaction 
indicators as part of a financial incentives program. 

Some MGD groups have chosen to undertake projects to improve satisfaction as a 
continuous quality improvement project. HPHC provides financial incentives and technical 
support to groups who choose to develop any continuous quality improvement activity 
including improving satisfaction. 



Health Economics Research, Inc. Access in Managed Care Plans: A-25 

crimson\final\apndx-a.wpd\dpb 



Appendix B 



Table B-l 



Sample Sizes for Access Indicators 



1. BREAST CANCER SCREENING RATE 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


Mammography 


Rate 


Sample 


Mammography 


Rate 


Sample 


Mammography 


Rate 


211,026 


86,112 


40.8% 


1,638 


1,312 


80.1% 


910 


608 


66.8% 


96,990 


52,128 


53.7% 


938 


812 


86.6% 


462 


361 


78.1% 


84,581 


30,384 


35.9% 


580 


444 


76.6% 


358 


219 


61.2% 


29,455 


3,600 


12.2% 


120 


56 


46.7% 


90 


28 


31.1% 



2. COLON CANCER SCREENING RATE 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


Screening Test 


Rate 


Sample 


Screening Test 


Rate 


Sample 


Screening Test 


Rate 


339,627 


121,508 


35.8% 


2,089 


1,274 


61.0% 


2,045 


1,106 


54.1% 


167,693 


63,934 


38.1% 


1,298 


819 


63.1% 


1,202 


691 


57.5% 


132,632 


47,582 


35.9% 


707 


427 


60.4% 


713 


370 


51.9% 


39,302 


9,992 


25.4% 


84 


28 


33.3% 


130 


45 


34.6% 



Pagel of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



3. RETINAL EXAMINATION RATE FOR DIABETES 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


Retinal Exam 


Rate 


Sample 


Retinal Exam 


Rate 


Sample 


Retinal Exam 


Rate 


34,260 


18,760 


54.8% 


1,092 


724 


66.3% 


495 


315 


63.6% 


16,925 


8,989 


53.1% 


745 


483 


64.8% 


305 


188 


61.6% 


14,214 


8,142 


57.3% 


322 


222 


68.9% 


161 


109 


67.7% 


3,121 


1,629 


52.2% 


25 


19 


76.0% 


29 


18 


62.1% 



4. VISIT RATE FOR DIABETES 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


two visits 


Rate 


Sample 


two visits 


Rate 


Sample 


two visits 


Rate 


34,260 


20,957 


61.2% 


1,092 


996 


91.2% 


495 


469 


94.7% 


16,925 


10,060 


59.4% 


745 


682 


91.5% 


305 


291 


95.4% 


14,214 


8,806 


62.0% 


322 


292 


90.7% 


161 


149 


92.5% 


3,121 


2,091 


67.0% 


25 


22 


88.0% 


29 


29 


100.0% 



Page 2 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



5. ADMISSION RATE FOR AMBULATORY CARE SENSITIVE CONDITIONS 
1994 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 



1995 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






HCD 






MGD 






Number of 


Rate 




Number of 


Rate 




Number of 


Rate 


Sample 


admissions 


(per 1000) 


Sample 


admissions 


(per 1000) 


Sample 


admissions 


(per 1000) 


363,481 


25,658 


70.59 


8,764 


410 


46.78 


4,196 


150 


35.75 


181,762 


9,033 


49.70 


5,829 


156 


27.28 


2,535 


71 


28.01 


140,436 


11,522 


82.04 


2,589 


183 


70.68 


1,389 


59 


42.48 


41,283 


5,103 


123.61 


346 


68 


196.53 


272 


20 


73.53 





Fee-for-Service 






HCD 






MGD 






Number of 


Rate 




Number of 


Rate 




Number of 


Rate 


Sample 


admissions 


(per 1000) 


Sample 


admissions 


(per 1000) 


Sample 


admissions 


(per 1000) 


325,984 


23,877 


73.25 


9,075 


420 


46.28 


4,319 


198 


45.84 


164,602 


8,214 


49.90 


5,769 


189 


32.76 


2,633 


88 


33.42 


126,472 


11,174 


88.35 


2,928 


167 


57.04 


1,394 


82 


58.82 


34,910 


4,489 


128.59 


378 


64 


169.31 


292 


28 


95.89 



Page 3 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



6. RATE OF PREHOSPITAL CARE FOR AMBULATORY CARE SENSITIVE ADMISSIONS 



1994 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


16,778 


13,839 


82.5% 


311 


261 


83.9% 


128 


107 


83.6% 


5,878 


4,788 


81.5% 


134 


117 


87.3% 


71 


60 


84.5% 


7,491 


6,292 


84.0% 


135 


114 


84.4% 


43 


36 


83.7% 


3,409 


2,759 


80.9% 


42 


30 


71.4% 


14 


11 


78.6% 



1995 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


22,577 


17,620 


78.0% 


310 


270 


87.1% 


153 


133 


86.9% 


7,325 


5,613 


76.6% 


159 


124 


78.0% 


83 


71 


85.5% 


10,477 


8,313 


79.3% 


116 


113 


97.4% 


54 


48 


88.9% 


4,775 


3,694 


77.4% 


35 


33 


94.3% 


16 


14 


87.5% 



Page 4 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



7. RATE OF POST-HOSPITAL CARE FOR AMBULATORY CARE SENSITIVE ADMISSIONS 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


A Visit 


Rate 


Sample 


A Visit 


Rate 


Sample 


A Visit 


Rate 


13,895 


11,210 


80.7% 


297 


245 


82.5% 


148 


126 


85.1% 


4,928 


4,056 


82.3% 


109 


96 


88.1% 


66 


61 


92.4% 


6,220 


5,079 


81.7% 


138 


116 


84.1% 


54 


43 


79.6% 


2,747 


2,075 


75.5% 


50 


33 


66.0% 


28 


22 


78.6% 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


A Visit 


Rate 


Sample 


A Visit 


Rate 


Sample 


A Visit 


Rate 


18,249 


13,891 


76.1% 


276 


229 


83.0% 


140 


120 


85.7% 


6,078 


4,744 


78.1% 


121 


106 


87.6% 


69 


60 


87.0% 


8,471 


6,474 


76.4% 


114 


94 


82.5% 


46 


40 


87.0% 


3,700 


2,673 


72.2% 


41 


29 


70.7% 


25 


20 


80.0% 



Page 5 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



8. ANTI-HYPERTENSIVE FOLLOW-UP RATE 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 




HCD 






MGD 




Sample 


Number with 

A Visit Rate 


Sample 


Number with 
A Visit 


Rate 


Sample 


Number with 
A Visit 


Rate 






3,078 


2,864 


93.0% 












1,622 


1,495 


92.2% 












1,179 


1,106 


93.8% 












277 


273 


98.6% 









9. ANTI-DEPRESSANT FOLLOW-UP RATE 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 




HCD 






MGD 




Sample 


Number with 

A Visit Rate 


Sample 


Number with 
A Visit 


Rate 


Sample 


Number with 
A Visit 


Rate 






1,121 


1,045 


93.2% 












627 


592 


94.4% 












407 


373 


91.6% 












87 


80 


92.0% 









Page 6 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



10. RATE OF POST-HOSPITAL FOLLOW-UP FOR MYOCARDIAL INFARCTION 



1994 







Fee-for-Service 






HCD 






MGD 








Number with 






Number with 






Number with 






Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


Total 


2,994 


2,222 


74.2% 


84 


76 


90.5% 


32 


30 


93.8% 


Age 65-74 


1,241 


970 


78.2% 


48 


44 


91.7% 


17 


16 


94.1% 


Age 75-84 


1,353 


1,007 


74.4% 


31 


28 


90.3% 


10 


9 


90.0% 


Age 85 and older 


400 


245 


61.2% 


5 


4 


80.0% 


5 


5 


100.0% 



1995 

Total 
Age 65-74 
Age 75-84 
Age 85 and older 



Fee-for-Service 





Number with 




Sample 


a visit 


Rate 


2,948 


2,138 


72.5% 


1,264 


973 


77.0% 


1,266 


917 


72.4% 


418 


248 


59.3% 



78 

33 
35 
10 



71 

32 
33 

7 



91.0% 
97.0% 
94.3% 
70.0% 



MGD 



Sample 



36 

24 

9 

3 



Number with 
a visit 



34 

23 
8 
3 



Rate 



94.4% 

95.8% 

88.9% 

100.0% 



Page 7 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



11. RATE OF POST-HOSPITAL FOLLOW-UP FOR DEPRESSION 
1994 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


963 


628 


65.2% 


12 


8 


66.7% 


12 


8 


66.7% 


364 


246 


67.6% 


9 


6 


66.7% 


8 


6 


75.0% 


472 


309 


65.5% 


3 


2 


66.7% 


4 


2 


50.0% 


127 


73 


57.5% 



















1995 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


Sample 


a visit 


Rate 


895 


595 


66.5% 


17 


12 


70.6% 


5 


5 


100.0% 


377 


258 


68.4% 


12 


11 


91.7% 


4 


4 


100.0% 


407 


261 


64.1% 


5 


2 


40.0% 


1 


1 


100.0% 


111 


76 


68.5% 



















Page 8 of 13 
crimson\final\tableBV B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



12. RATE OF FOLLOW-UP FOR ABNORMAL MAMMOGRAM 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 






HCD 






MGD 






Number with 






Number with 






Number with 




Sample 


Follow-up 


Rate 


Sample 


Follow-up 


Rate 


Sample 


Follow-up 


Rate 








76 


67 


88.2% 














45 


42 


93.3% 














28 


22 


78.6% 














3 


3 


100.0% 









13. RATE OF BREAST CANCER ONCOLOGY FOLLOW-UP 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 






HCD 






MGD 




Sample 


Number with 
a Visit 


Rate 


Sample 


Number with 
a Visit 


Rate 


Sample 


Number with 
a Visit 


Rate 








162 


115 


71.0% 














90 


69 


76.7% 














55 


39 


70.9% 














17 


7 


41.2% 









Page 9 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



14. POPULATION BASED RATE OF LENS REPLACEMENT 
1994 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


363,481 


13,482 


37.09 


181,762 


4,632 


25.48 


140,436 


6,911 


49.21 


41,283 


1,939 


46.97 





HCD 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


8,764 


231 


26.36 


5,829 


106 


18.18 


2,589 


104 


40.17 


346 


21 


60.69 





MGD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


4,196 


75 


17.87 


2,535 


26 


10.26 


1,389 


37 


26.64 


272 


12 


44.12 



1995 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


325,984 


12,623 


38.72 


164,602 


4,743 


28.81 


126,472 


6,404 


50.64 


34,910 


1,476 


42.28 





HCD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


9,075 


260 


28.65 


5,769 


119 


20.63 


2,928 


114 


38.93 


378 


27 


71.43 





MGD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


4,319 


67 


15.51 


2,633 


27 


10.25 


1,394 


30 


21.52 


292 


10 


34.25 



Page 10 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



15. POPULATION BASED RATE OF HIP AND KNEE REPLACEMENT 



1994 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 

1995 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


363,481 


2,393 


6.58 


181,762 


1,203 


6.62 


140,436 


1,058 


7.53 


41,283 


132 


3.20 





HCD 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


8,764 


53 


6.05 


5,829 


34 


5.99 


2,589 


18 


6.95 


346 


1 


2.89 





Fee-for-Service 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


325,984 


2,258 


6.93 


164,602 


1,226 


7.45 


126,472 


906 


7.16 


34,910 


126 


3.61 





HCD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


9,075 


50 


5.51 


5,769 


28 


4.85 


2,928 


19 


6.49 


378 


3 


7.94 





MGD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


4,196 


30 


7.15 


2,535 


28 


7.49 


1,389 


10 


7.20 


272 


2 


7.35 





MGD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


4,319 


37 


8.57 


2,633 


22 


7.49 


1,394 


14 


10.04 


292 


1 


3.42 



Page 11 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



16. POPULATION BASED RATE OF CORONARY REVASCULARIZATION 



1994 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 

1995 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-for-Service 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


363,481 


3,027 


8.33 


181,762 


1,889 


10.39 


140,436 


1,064 


7.58 


41,283 


74 


1.79 





HCD 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


8,764 


74 


8.44 


5,829 


53 


9.09 


2,589 


21 


8.11 


346 





0.00 





Fee-for-Service 






Number with 


Rate 


Sample 


a procedure 


(per 1000) 


325,984 


2,894 


8.88 


164,602 


1,820 


11.06 


126,472 


1,011 


7.99 


34,910 


63 


1.80 





MGD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


4,196 


20 


4.77 


2,535 


16 


6.31 


1,389 


4 


2.88 


272 





0.00 





HCD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


9,075 


67 


7.38 


5,769 


36 


6.24 


2,928 


30 


10.25 


378 


1 


2.65 





MGD 






Number with 


Rate 


Sample 


a procedure 


(perlOOO) 


4,319 


19 


4.40 


2,633 


16 


6.08 


1,394 


3 


2.15 


292 





0.00 



Page 12 of 13 
crimson\final\tableB\ B-1 \ss 



Table B-l 



Sample Sizes for Access Indicators (continued) 



17. NEW ENROLLEE VISIT RATE 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 




HCD 






MGD 






Number with 




Number with 






Number with 




Sample 


A Visit Rate 


Sample 


A Visit 


Rate 


Sample 


A Visit 


Rate 






750 


554 


73.9% 


750 


365 


48.7% 






464 


346 


74.6% 


440 


194 


44.1% 






224 


163 


72.8% 


248 


133 


53.6% 






62 


45 


72.6% 


62 


38 


61.3% 



18. ANNUAL VISIT RATE 



Total 
Age 65-74 
Age 75-84 
Age 85 and older 





Fee-For-Service 






Number with 




Sample 


A Visit 


Rate 


325,984 


288,083 


88.4% 


164,602 


141,602 


86.0% 


126,472 


114,562 


90.6% 


34,910 


31,919 


91.4% 



Sample 



HCD 



500 
281 
178 

41 



Number with 
A Visit 



Rate 



470 
267 
163 

40 



94.0% 
95.0% 
91.6% 
97.6% 



Sample 



MGD 



500 
261 
179 

60 



Number with 
A Visit 



454 
234 
166 

54 



Rate 



90.8% 
89.7% 
92.7% 
90.0% 



Page 13 of 13 
crimson\final\tableB\ B-1 \ss 



Appendix C 



HPHC Enrollment File Layout 



•EMOGRAPHIC DOWNLOAD DATA DICTIONARY - OUTPUT ORDER 



FIELD 
NAME 



SIZE 



POSITION TYPE 



DESCRIPTION 



ENCNTR 


1 


1 


C 


"D" FOR DEMOGRAPHIC 


TAPEID 


3 


2-4 


C 


TAPE ID # 


INPUTMRN 


8 


5-12 


C 


UNIT # FROM INPUT LIST 


CURRMRN 


8 


13-20 


C 


CURRENT UNIT # 


FAMNUM 


9 


21-29 


C 


FAMILY n 


DOB 


8 


30-37 


N 


DATE OF BIRTH 


ZIPCODE 


5 


38-42 


C 


ZIP CODE 


ORGEFF 


6 


43-48 


N 


ORIGINAL EFFECTIVE DATE 


PREEFF 


6 


49-54 


N 


EFFECTIVE DATE PRECEEDING TIME WINDOW 


PREGRP 


6 


55-60 


C 


GROUP CODE PRECEEDING TIME WINDOW 


PREDIV 


3 


61-63 


C 


DIVISION CODE PRECEEDING TIME WINDOW 


PREDUAL 


3 


64-66 


c 


DUAL DIVISION CODE PRECEDING TIME WINDOW 


PREBEN 


4 


67-70 


c 


BENEFIT PACKAGE (COVERAGE CODE) PRECEEDING TW 


PRECOV 


4 


71-74 


c 


COVERAGE MODIFIER PRECEEDING TIME WINDOW 


PREDRUG 


1 


75 


c 


PRE-PAID DRUG BENEFIT (Y/N) PRECEEDING TW 


PREDEP 


2 


76-77 


c 


DEPENDENCY CODE PRECEEDING TIME WINDOW 


PRECONT 


1 


78 


c 


CONTRACT TYPE PRECEEDING TIME WINDOW 


EFFDATE 


6 


79-84 


N 


EFFECTIVE DATE IN TIME WINDOW 


GRPCODE 


6 


85-90 


C 


GROUP CODE IN TIME WINDOW 


VISION 


3 


91-93 


C 


DIVISION CODE IN TIME WINDOW 


jUALDIV 


3 


94-96 


C 


DUAL DIVISION CODE IN TIME WINDOW 


OVCODE 


4 


97-100 


C 


BENEFIT PACKAGE (COVERAGE CODE) IN TW 


COVMOD 


4 


101-104 


c 


COVERAGE MODIFIER IN TIME WINDOW 


DRUGBEN 


1 


105 


c 


PRE-PAID DRUG BENEFIT (Y/N) IN TIME WINDOW 


DEPCODE 


2 


106-107 


c 


DEPENDENCY CODE IN TIME WINDOW 


CONTRACT 


1 


108 


c 


CONTRACT TYPE IN TIME WINDOW 


CONTMON 


6 


109-114 


N 


CONTINUOUS MONTHS UP TO TIME WINDOW 


MEMBMON 


6 


115-120 


N 


MEMBER-MONTHS IN TIME WINDOW 


GENDER 


1 


121 


r 


MEMBER'S SEX 


RACE 


2 


122-123 


C 


MEMBER'S RACE CODE(S) 


PMD 


4 


124-127 


C 


PRIMARY MD 


PNP 


4 


128-131 


C 


PRIMARY NP 


HOSP 


3 


132-134 


N 


# OF HOSPITALIZATIONS ARCHIVED 


TOTHOSP 


3 


135-137 


N 


TOTAL ft OF HOSPITALIZATIONS 


OLDREC 


8 


138-145 


C ^ 


ORIGINAL RECORD NUMBER 



HCD Internal Medicine Encounter Form 



HCHP INTERNAL MEDICINE ENCOUNTER FORM 



4) SITE OF ENCOUNTER 5) TYPE OF ENCOUNTER 



DC. 

BA. 

BUR. 

B. 

CH. 
CO. 

C 
MA. 
PA. 
QU. 
SV. 
WA. 

W. 
WR. 

H . 

Y. 

E . 

J. 

K . 



. BOSTON 
. BRAINTREE 
. BURLINGTON 
. CAMBRIDGE 
. CHELMSFORD 
. COPLEY 
. KENMORE 
. MEDFORD 

PEABODY 

QUINCY 
. SOMERVILLE 
. WATERTOWN 

WELLESLEY 

WEST ROXBURY 

Bl 

B/W 

CHMC 

HOUSE CALL 

OTHER: 



A 


srHFDi 'LED 


R 


SAME DAY 


C 


TELEPHONE 


W 


DNK/CANCELLED 


G 


IN-PATIENT 


H 


EW 


1 


NON-ENCOUNTER 


O 


LETTER SENT 



NAME: 

UNIT#: 
DOB: 

DATE: 

PROV: 



CODE: 



M. 



HOSPITAL DISCHARGE SUMMARY: specify 

Primary diagnosis for input: 

Admission date: L I 

6) EW or Hosp. Visit Approved? Yes . 



.No 



PLEASE USE RED INK 



PERSONAL BACKGROUND/DEMOGRAPHIC 



PLEASE USE RED INK 



FOR USE BY PRIMARY PROVIDER ONLY-Any previous entry will be overwritten if new information is indicated in any of the fields below 

7) PRIMARY MD: 9) PRIMARY RN: 

9) RACE A _ 



Caucasian C 
Black D 

E 



11) NO. CHILDREN: 

12) OCCUPATION: 

38) PERSONAL BACKGROUND 



Spanish-Speaking 

Other 

Asian 



10) MARITAL STATUS: A 
B 
C 



Single D . 

Married E 

Widowed F 



39) EMERGENCY TELEPHONE NO 



Separated 

Divorced 

Cohab 



ADMINISTRATIVE DATA 

40) DISPOSITION [choose one or more - free text allowed with each choice) 

Return visit with in 

E _ 
F 



I 



Patient to call MD 


G 


Patient to call RN 


MD to call Patient 


H 


RN to call Patient 


Other: 







A 
B 
C 
D 



Days 
Weeks 
Months 
PRN 



42) INTERNAL HCHP CONSULTATION(S) 



Referred to: 

Specialty (REQUIRED): _ 

Provider (optional): 

DX Code(s) (REQUIRED) 



Referred to: 

Specialty (REQUIRED): 

Provider (optional): 



DX Code(s) (REQUIRED): 



92) SEND CONSULTATION SUMMARY TO: ,S,mple check sends to pnmary provider) 

It summary is to be sent to additional provider please indicate 

Prov.: Prov. Code: 

Prov.: Prov. Code: 



149) Owns TTY: Telephone for the deaf Yes 

46) DX codes for future extended output: 

47) Review of encounter (7 'o 10 days) 48) _ 

54) Paper chart required at next visit 



Discontinued 



. Long-term follow-up important (90 days) 



64) BLOOD PRESSURE 
RIGHT 

Lyng 

Sitting 

Standing __ 



9CK tg cjft i* indicated! 

LEFT 
lg cuff 



OBJECTIVE DATA 60) height 

61) WEIGHT 

62) PULSE 



lbs. 



/mm. 



Jg cuff 



68) RESP. RATE. 
63) TEMP. 

1)oral 



_2) rect. 



_3) axil. 



MASTER SYNONYM LIST 

TO ADD MASTER SYNONYM LIST :To /om together exstmg codes wh'Cn reter to the same probiemi 
Master Code Problem name 



Subsidiary Synonymis) Code_ 

(maximum of five) Code_ 

Code. 

Code, 

Code 



Problem name_ 
Problem name_ 
Problem name_ 
Problem name. 
Problem name 



TO KILL EXISTING MASTER SYNONYM T'M.»si,n 9 '/sti Master Code. 



1/94 



Problem Name. 



PLEASE USE RED INK 



DIAGNOSES AND PROBLEMS (Dx) 



PLEASE USE RED INK 



M = Major, = Omit from Status Report, P = Presumptive, S/P = Status Post, R/O = Rule Out 
I = Inactive, H = History of, /= Minor 



A600 IHA 



A800 PHR 



1. skmnl 2. eyes nl 3. ENT nl 4. thyrnl 



5. lungs nl 



8. abd nl 9. breast nl 10. GU nl 



11. rectal nl 12. musc/skel nl 



6. cardiac nl 



13. neuro nl 



7. vase nl 



19. nodes nl 



HEALTH HABITS (Providers: Indicate Smoking 1 nerapies on Pg 4) 

NOTES: . 





A062 




A063 




A064 




P106 




T050 




D307 




A178 




A014 



Current Smoker (Frequency) 

Never Smoker 

Former Smoker; Yrs since quitting: 

Alcohol Use (Frequency) 

Exercise (Frequency) 

Seatbelt Use (Frequency) 

Case Reviewed With Dr.: 



ENT 



GENERAL 





A020 




A990 




A802 




A810 




A177 




Q137 




A003 




A801 




A019 




A128 




A803 



Abnormal Physical Finding (Specify/ 



Abnormal test result 

Dx Deferred 

Exam for certificate 

Health education 

HIV health education 

HIV Testing Performed 

Immunization 

No demonstrable disease (explain) 

Positive family Hx (specify) 

Rx refill only 
Test results only 





E100 




E230 




E173 




E130 




E120 




E153 




E154 




E410 




E408 




E250 




E252 




E260 




E401 



THYROID 



DRUG REA CTIONS (Specify All Drugs & Reactions) 





B210 




B151 




B152 




B153 



Cerumenosis (earwax) 

Epistaxis 

Hearing loss 

Labyrinthitis 

Otitis externa 

Otitis media, serous 

Otitis media, suppurative 

Phar,..gitis 

Pharyngitis, strep (by culture) 

Rhinitis 

Rhinitis, allergic 

Sinusitis 

Tonsillitis 



Goiter 

Hyperthyroidism 
Hypothyroidism 
Thyroid nodule(s) 



A145 



SYSTEMIC 



Drug allergy status 

2) None known 
Drug intolerance 



Fatigue 

Viral illness 
Weight loss 



TRAUMA (STATE SITE IN FREE TEXT» 
Contusion 
Burn 

Head Trauma 
Laceration 
Puncture wound 
Trauma other 





A991 




A117 




A992 





B120 




N011 




B160 




B005 




L180 




A150 





N045 




C132 




O300 




C270 




C484 




N014 



SKIN 





G100 




G121 




G122 




G261 




G992l 




G993 




G220 




G249 




G270 





C408 




S420 




C260 




C150 




C162 




C160 




C183 




C220 




C230 




C306 




C330 




C337 




C492 




C340 




C165 




C430 




C166 




C453 




C191 




C375 




C370 




C441 




C440 | 



EYE 



| D410 | 


D525 




D431 




D007 




D330 




D450 




D114 



Abscess 

Acne vuigans 

Actinic keratosis (Senilis) 

Cellulitis 

Contact dermatitis 

Dermatitis lunknown etiology) 

Eczema 

Herpes s.mp.ex 

Herpes zoster ishmgiesi 

L.poma 

Nevus 

Paronychia 

P-untuS 

Psor as>S 

Sebormeic dermatitis 

Seoorme'C keratosis 

Stasis dermat'tis 

Tnea cruns 

T nea peais 'Athlete s ^ooti 

T.nea versicolor 

Urticaria 

J&'^ca pianta ri S i Plantar Wartsi 

/e"i>ca /, : aa f s iWansi 



Ca'.a'ac* 
Control ' 



Comeai acasion 
Foreign Dody - eye 
Glaucoma 
Iritis 

S:ve iHoraeonj"-! 



ENDOCRINE METABOLIC 

Diabetes Mellitus 

Gout 

Hypercholesterolemia 

Hyperlipoproteinemia 

Impotence 

Obesity 

RESPIRATORY 

Asthma 

Bronchitis, acute 

Bronchitis, chronic 

COLD. 

Cough 

Dyspnea (shortness of breath) 

Pneumonia (state variety) 

Positive PPD 
URI 

/ASCULAR 

Aortic Stenosis 

Arrhythmia 

Atrial fibrillation 

Chest pain 

Congestive heart failure 

Coronary artery disease 

Ectopic beats 

Hypertension 

Mitral valve prolapse 

Myocardial intarction 

Murmur, systolic 

Palpitations 

Peripheral Vascular Disease 

Tachycardia 

Thrombophlebitis 

Varicose veins 

ILOGY / A.I.D.S. 

Acquired Immune Deficiency Syndrome (A.I D.S ) 

AIDS related complex (A.R C ) 

Anemia 

HIV infection 

Infectious mononucleosis 

Iron deficiency anemia 

Leukopenia 

lymphadenopathy 

Sickle cell trait 

Thalassemia trait 

Thrombocytopenia 





1120 




1136 




1131 




G991 




1160 




I099 




1137 




J110 




1971 




I270 




1211 




1102 




Q175 




1105 




J130 




J140 





Q10B 




Q136 




Q110 




Q135 




Q140 




0118 




Q183 




O302 




Q131 




Q133 




Q203 



K103 



S/P 



K261 



S/P 



K403 



K132 



K311 



S271 



K340 



K150 



K161 



K160 



K99B 



K180 



K110 



K210 



K220 



K114 



K230 



K270 



Alccno lic hepatitis 



Y459 | Appendectomy 



Cholec ystitis 



Y535 I Cholecystectomy 



Cholelithiasis 

Colitis, ulcerative 

Colon Polyp 

Constipation 

Crohn's Disease (Regional Ententis) 

Diarrhea 

Diverticulitis 

Diverticulosis 

Dyspepsia 

Esophagitis 

Functional Gl complaints 

Gastritis 

Gastroenteritis 

Gl bleeding 

Hemorrhoids 

Hepatitis 1) Type A 2)TypeB_ 





L140 




L993 




L190 




L11S 




L222 




B170 




L230 




L101 



VENEREAL 





M172 




L160 




C235 




L290 




L291 




L250 



_3)Type C 





K241 




K401 




K136 




K280 




K11S 




K910 



Esophageal (Hiatus) hernia 

Inguinal hernia 

Irritable 8owel Syndrome 

Pancreatitis 

Peptic ulcer disease 

Normal rectal exam 1 ) Hemoccult cards given 



K113 I Rectal bleeding 





Y401(R) 




W013(T) 




Y150(T) 



Anoscopy 

Sigmoidoscopy (Informed consent may be required) 

Stool hematest Pos Neg 



BREAST 





H190 




H107 




H112 




H111 




H109 




H114 




H113 




H108 




H106 



Normal breast exam 1) SBE Taught 

Fibrocystic breast changes 
Breast lump: upper, inner quadrant 1 ) size 
Breast lump: upper, outer quadrant 1) size 
Breast lump: lower, inner quadrant 1 ) size 
Breast lump: lower, outer quadrant 1 ) size 

Breast lump: Other (specify site) 

Mammogram Ordered 
Abnormal Mammogram 

GYNECOLO GY 

I I M137~l Routine GYN exam _ 
3) vag nl 4) adnexa nl 



2) SBE Reviewed 



_cm 
_cm 

_cm 



1) ext. genitalia nl 2) cervix nl 

. 5) Pap done 6) uterus nl 



M019 



M124 



M135 



M130 



M206 



M320 



M116 



M133 



S/P 



Abnormal Pap Smear 

Amenorrhea 

Cervicitis 

Dysfunctional bleeding (metrorrhagia) 

Dysmenorrhea 

Endometriosis 

Family planning 

Fibroids, Uterine 

Hyperm enorrhea (Menorrhagia) 

Y720 | Hysterectomy 



Epididymitis 

Hematuria 

Kidney stones (Nephrolithiasis) 

Normal testicular exam 1 ) TSE taught 

Prostatitis 

Proteinuria 

Pyelonephritis 

Urinary Tract Infection (UTI) 

Condyloma acuminatum 

Gonorrhea 

Herpes progenitalis 

Urethritis 

Urethritis. Non-specific 

Syphilis 

MUSCULOSKELETAL 

Arthritis 

Bursitis 

Carpal Tunnel Syndrome 

Cervical radiculitis 

Disc disease, cervical 

Disc disease, lumbar 

Fracture 

Ganglion 

Joint effusion 

Joint pain 

Low back pain 

Musculo -skeletal pain 

Osteoarthritis 

Osteoporosis 

Plantar Fasciitis 

Rheumatoid arthritis 

Sprain 

Tendonitis 
NEUROLOGIC 

Cerebrovascular Accident (CVA) 

Dizziness 

Headache (undifferentiated) 

Migraine headache 

Seizures 

Tension headache 

Transient Ischemic Attack (TIA) 
PSYCHOSOCIAL 

Alcoholism 

Anxiety 

Depression 

Substance abuse/dependence 

Problem of living 

Acute situational disturbance 

Chronic situational disturbance 





N010 




N020 




N066 




N033 




N061 




N261 




N090 




N476 




N145 




N108 




N992 




N987 




N013 




N140 




N082 




N016 




N170 




N181 





01 50 




E125 




0991 




01 90 




0124 




0191 




0127 





P100 




P011 




P120 




P549 




P600 




P302 




P303 



CANCER 



M995 I Infertility 

M140 | Menopausal Syndrome 

1) Hormone replacement therapy: Risk/Benefits Discussed 

M136 Oligom enorrhea 
S/P | Y701~ l Oophorectomy 





H100 




H101 




K123 




G130 




L221 



Carcinoma (specify)_ 
Carcinoma of breast 
Carcinoma of colon 
Cancer lung 
Cancer of prostate 



M150 



M175 



M243 



M200 



M160 



S/P 



M996 



|B236(T) 



Ovarian cyst 

Pelvic Inflammatory Disease 

Pelvic mass 

Pelvic pain 

Pregnancy 

Preme nstrual tension/syndrome 

Y372 | Tubal ligation 

Vaginitis 1 ) Whiff Test 2) pH 

Wet Prep Pos Neg 



FREE TEXT COMMENTS RE: DIAGNOSES & PROBLEMS 

Code # Text 



PRINCIPAL DIAGNOSIS FOR THIS ENCOUNTER 

The Principal Diagnosis listed in this section should also be listed in the Diagnosis section ot this form 
(DO NOT WRITE ANY FREE TEXT IN THIS SECTION) 



PRINCIPAL DIAGNOSIS: 



iDX Text] 



PLEASE USE RED INK 



Ul-l-ICt CONsULl AIlONS/UrrlCfc Vial i 6 t.RA) 
Please do not free text on Consultation and Visit Codes 



PLEASE USE RED INK 



NEW PT. VISIT 

L1 (Prob foe. HX & Exam. Strtfrwd Dec) 
L2 (Expand Prob HX & Exam, Strtfrwd Oec) 
L3 (Detailed HX & Exam, Low Complex Dec) 
L4 (Compre HX & Exam, Mod Complex Dec) 
L5 (Compre HX & Exam, High Complex Dec) 





T335 




T336 




T337 




T338 




T339 



Consultation and Visit Codes will be suppressed on Standard Record Summaries. 
ESTABLISHED PT. VISIT 

L1 (Minimal Prob, May Not Req MD) 





T340 




T341 




T342 




T343 




T344 



L2 (Prob foe HX & Exam, Strtfrwd Dec) 
L3 (Expand Prob HX & Exam, Strtfrwd Dec) 
L4 (Detailed HX & Exam, Mod Complex Dec) 
L5 (Compre HX & Exam, High Complex Dec) 



PHONE CONSULT/MED MG! 

Simple/Brief 
Intermediate 
Lengthy/Complex 

TIVE COUNSELING 

1 5 Minutes 
30 Minutes 





T462 




T463 




T464 





T012 




T013 



THERAPIES (Rx) 

/ = Active I = Inactive O = Omit from Status Report and Standard Record Summary 

PULMONARY FUNCTION TEST (Rx) (Inputters: Please include "Liters/min .' as free text ) 



I I Y155~l PEAK FLOW . 



Liters/min. 



A) 1 00% expected value for age, height and sex 

B) Patient's best ever 
I T900 J Oxygen Saturation, By Oximetry 



C) Pretreatment 

D) Post three sympathomimetic treatments 



PRESCRIPTIONS 





D107 




H147 




Dill 




D160 




F126 




I237 




I238 




FI32 




1122 





D180 




D143 




D144 




E142 




E171 




1111 




D152 




H206 




J133 




J109 




1355 




D167 




C119 



Atenolol (Tenormin) 

Cimetidine (Tagamet) 

Digoxin (Lanoxin) 

Diltiazem (Cardizem) 

Furosemide (Lasix) 

Glipizide (Glucotrol) 

Glyburide (Micronase, Diabeta) 

Hydrochlorothiazide (Hydrodiuril) (HCTZ) 

Insulin NPH 

Regular 

Lente 

Lismopril (Zestnl, Pnnivil) 
Lopressor, (Metoprolol) 
Nifedipine (Procardia, Adalat) 
Phenobarbital 
Phenytoin (Dilantin) 
Prednisone 
Propranolol (Inderal) 
Ranitidine (Zantac) 
Theodur 

Theophylline Prep (specify) 

Tolazamide (Tolinase) 
Verapamil (Calan, Isoptin) 
Warfann Sodium (Coumadin) 





E116 




G115 




P143 




H110 




E111 




F149 




M272 



OVERT E COUNTER 

Acetaminophen (Tylenol) 

Actifed 

Afrin 

Antacids (specify) 

Aspirin 

Calcium Carbonate 

Clotrimazole 

1 topical crm (Lotrimin) _ 

3 vag crm (Gyne-Lotnmin) _ 

Diphenhydramine HCI (Benadryl) 

Ibuprofen (Advil, Motrin) 

Meclizine (Antivert) 

Metamucil 

Pseudoephednne (Sudafed) 

Tolnaftate (Tinactin) 
OMATIC, Etc. 

Application of cold locally 

Application of heat locally 

Collar 

Counseling 

Crutches 

Diet Counseling 

Gargle 

Lozenges, sore throat 

Smoking Counseling 

Smoking: Referred to program 

Soaks 

Symptomatic Rx (specify) 





A110 




E098 




A132 




H133 




J110 




M139 



_2 topical sol (Lotrimin) 
_4 vag tab (Gyne-Lotnmin 



T653 



T654 



S101 



R305 



R077 



V121 



R014 



M992 



R021 



R0022 



R028 



R001 



PRESCRIPTION INFORMATION 



CODE 


DRUG 


STRENGTH 


QUAN 


# REFILLS 


DIRECTIONS 







































r 



PLACE Rx PAD STICKERS HERE (Do not cover hand-written text) 



~l 



Rx 1 



h 



H 



Rx2 



h 



H 



Rx3 





Q107 




Q186 




Q160 




CI101 




Q130 




Q156 




O060 




O201 



L 



Haemophilus Vac (B-Capsa 

MMR (1) 

Rubella Vaccine 

Td (Adult) 

Tetanus Toxoid 

Hepatitis B Vac (Recombivax-HB) 

Flu Vaccine 

Gamn-a Globulin fimm-jne S?' 



J 



IMMUNIZATIONS AND TESTS (Rx) 





MFor H 


Lot or Date 


VAD 


I 











































0157 




Q200 




Q185 




Q301 




Q302 




Q055 




Q190 







Hepatitis B Vac. (Engenx-B) 
Cholera Vaccine 
Pneumococcal Vaccine 
PPD 1st Strenglh-1TU 
PPD Int. Strength-5TU 
Tine Test 
Typhoid Vaccination 



IMP 


POS 


NEG 





















HCD Encounter Data File Layout 



ENCOUNTER DOWNLOAD DATA DICTIONARY - OUTPUT ORDER 



FIELD 
NAME 



SIZE 



POSITION 



TYPE 



DESCRIPTION 



ENCNTR 


1 


1 


C 


"E" FOR ENCOUNTER, "F" FOR CONTINUATION 


TAPEID 


3 


2-4 


C 


TAPE ID # 


INPUTMRN 


8 


5-12 


c 


UNIT tt FROM INPUT LIST 


CURRMRN 


8 


13-20 


c 


CURRENT UNIT # 


VISDATE 


6 


21-26 


N 


VISIT DATE 


VISPROV 


4 


27-30 


c 


VISIT PROVIDER #1 


SPEC 


5 


31-35 


c 


SPECIALTY i 


VISPROV2 


4 


36-39 


c 


VISIT PROVIDER #2 


SPEC2 


5 


40-44 


c 


SPECIALTY 


SITE 


3 


45-47 


c 


SITE 


TYPE 


1 


48 


c 


TYPE 


OLDREC 


8 


49-56 


c 


OLDREC - ORIGINAL RECORD NUMBER 


DOB 


8 


57-64 


N 


DATE OF BIRTH 


GENDER 


1 


65 


C 


MEMBER'S SEX 


RACE 


2 


66-67 


c 


MEMBER'S RACE CODE(S) 


PMD 


4 


68-71 


c 


PRIMARY MD 


ENCMRN 


8 


72-79 


c 


ENCMRN - MRN ON DATE OF ENCOUNTER 


ENCHC 


2 


80-81 


c 


ENCHC - HC ON DATE OF ENCOUNTER 


DCODE1 


4 


82-85 


c 


D-CODE 1 


DSTAT1 


1 


86 


c 


STATUS 


DCODE2 


4 


87-90 


c 


D-CODE 2 


DSTAT2 


1 


91 


c 


STATUS 


DCODE3 


4 


92-95 


c 


D-CODE 3 


DSTAT3 


1 


96 


c 


STATUS 


DCODE4 


4 


97-100 


c 


D-CODE 4 


DSTAT4 


1 


101 


c 


STATUS 


DCODE5 


4 


102-105 


c 


D-CODE 5 


DSTAT5 


1 


106 


c 


STATUS 


DCODE6 


4 


107-110 


c 


D-CODE 6 


DSTAT6 


1 


111 


c 


STATUS 


DCODE7 


4 


112-115 


c 


D-CODE 7 


DSTAT7 


1 


116 


c 


STATUS 


DCODE8 


4 


117-120 


c 


D-CODE 8 


DSTAT8 


1 


121 


c 


STATUS 


DCODE9 


4 


122-125 


c 


D-CODE 9 


DSTAT9 


1 


126 


c 


STATUS 


DCODEIO 


4 


127-130 


c 


D-CODE 10 


DSTAT10 


1 


131 


c 


STATUS 


NDCODES 


3 


132-134 


c 


TOTAL D-CODES 


TCODE1 


4 


135-138 


c 


T-CODE 1 


TSTAT1 


1 


139 


c 


STATUS 


TCODE2 


4 


140-143 


c 


T-CODE 2 


TSTAT2 


1 


144 


c 


STATUS 


TCODE3 


4 


145-148 


c 


T-CODE 3 


TSTAT3 


1 


149 


c 


STATUS 


TCODE4 


4 


150-153 


c 


T-CODE 4 


TSTAT4 


1 


154 


c 


STATUS 



TC0DE5 


4 


155-158 


C 


T-CODE 5 


TSTAT5 


1 


159 


C 


STATUS 


TC0DE6 


4 


160-163 


C 


T-CODE 6 


TSTAT6 


1 


164 


C 


STATUS 


TC0DE7 


4 


165-168 


C 


T-CODE 7 


TSTAT7 


1 


169 


C 


STATUS 


TC0DE8 


4 


170-173 


C 


T-CODE 8 


TSTAT8 


1 


174 


C 


STATUS 


TC0DE9 


4 


175-178 


C 


T-CODE 9 


TSTAT9 


1 


179 


c 


STATUS 


TCODE10 


4 


180-183 


c 


T-CODE 10 


TSTAT10 


1 


184 


c 


STATUS 


TCODE1 1 


4 


185-188 


c 


T CODE 1 1 


TSTAT1 1 


1 


189 


c 


STATUS 


TC0DE12 


4 


190-193 


c 


T CODE 12 


TSTAT12 


1 


194 


c 


STATUS 


TC0DE13 


4 


195-198 


c 


T CODE 13 


TSTAT13 


1 


199 


c 


STATUS 


TC0DE14 


4 


200-203 


c 


T CODE 14 


TSTAT14 


1 


204 


c 


STATUS 


TC0DE15 


4 


205-208 


c 


T CODE 15 


TSTAT15 


1 


209 


c 


STATUS 


TC0DE1 6 


4 


210-213 


c 


T CODE 16 


TSTAT16 


1 


214 


c 


STATUS 


TC0DE17 


4 


215-218 


c 


T CODE 17 


TSTAT1 7 


1 


219 


c 


STATUS 


TC0DE18 


4 


220-223 


c 


T CODE 18 


TSTAT18 


1 


224 


c 


STATUS 


TC0DE19 


4 


225-228 


c 


T CODE 19 


TSTAT19 


1 


229 


c 


STATUS 


TCODE20 


4 


230-233 


c 


T CODE 20 


TSTAT20 


1 


234 


c 


STATUS 


NTCODES 


3 


235-237 


c 


TOTAL T-CODES 


RCODE1 


4 


238-241 


c 


R-CODE 1 


RSTAT1 


1 


242 


c 


STATUS 


RCODE2 


4 


243-246 


c 


R-CODE 2 


RSTAT2 


1 


247 


c 


STATUS 


RCODE3 


4 


248-251 


c 


R-CODE 3 


RSTAT3 


1 


252 


c 


STATUS 


RCODE4 


4 


253-256 


c 


R-CODE 4 


RSTAT4 


1 


257 


c 


STATUS 


RCODE5 


4 


258-261 


c 


R-CODE 5 


RSTAT5 


1 


262 


c 


STATUS 


RCODE6 


4 


263-266 


c 


R-CODE 6 


RSTAT6 


1 


267 


c 


STATUS 


RCODE7 


4 


268-271 


c 


R-CODE 7 


RSTAT7 


1 


272 


c 


STATUS 


RCODE8 


4 


273-276 


c 


R-CODE 8 


RSTAT8 


1 


277 


c 


STATUS 


RCODE9 


4 


278-281 


c 


R-CODE 9 


RSTAT9 


1 


282 


c 


STATUS 


RCODE10 


4 


283-286 


c 


R-CODE 10 



RSTAT10 


1 


287 


C 


STATUS 


NRCODES 


3 


288-290 


C 


TOTAL R-CODES 


MHVIST 


1 


291 


C 


MENTAL HEALTH VISIT (A75) 


SAVIST 


1 


292 


C 


SUBSTANCE ABUSE VISIT <A76> 


WEIGHT 


4 


293-296 


C 


WEIGHT 


TEMP 


6 


297-302 


C 


TEMP 


BP 


6 


303-308 


C 


BP 


PATCODE 


8 


309-316 


C 


PAT CODE 


GRPCODE 


6 


317-322 


C 


GROUP CODE 


DIVCODE 


3 


323-325 


C 


DIVISION CODE 


CENCODE 


4 


326-329 


C 


BENEFIT PACKAGE (COVERAGE CODE) 


COVMOD 


4 


330-333 


C 


COVERAGE MODIFIER 


PPDRUG 


1 


334 


C 


PRE-PAID DRUG BENEFIT (Y/N) 



HCD Outside Utilization File Layout 



Chapter 6: Data Dictionary 



UEUSN Data Dictionary 



Introduction 



This section contains the UEUSN Data Dictionary. The columns in the dictionary 
have the following meanings: 

• UEUSN field name - This is the name of the field in UEUSN. (Fields names 
remain constant across the Universal Extract and the health center files.) 
These are the names you use when creating reports in Decision Analyzer. 

• Report header name - This column serves two purposes: 

1: It helps describe the UEUSN field name by linkin g it to the actual 

screen name. 
2: It is a name you can use as a heading in a printed report. 

• Size - The length of the field. (This is the number of spaces the field takes up 
in the database and not necessarily the size of the field as it appears on the 
screen.) 

• Type - C indicates the field is a character field. N is a numeric field. Only N 
fields can be used to calculate new fields in Decision Analyzer reports. There 
are fields that contain numbers, (e.g., contract number), that are defined as 
character fields because they cannot be used in numeric calculations. 

• Description - This is a short definition of the field. Some fields include 
additional information, such as valid values. 



Note: Fields preceded by a C are specific to a claim; those preceded by a P are referral 
related. (There are exceptions to this rule.) 



Data Dictionary 



The UEUSN Data Dictionary — sorted alphabetically by UEUSN field name — is 
shown below: 



UEUSN Report c . _ 
_. .... u j m Size T yP e Description 
Field Name Header Name < 


AUTH 


Auth# 


12 


C 


The referral number. Referral numbers consist 
of the referral date, site code, and a sequential 
number. 


CADMDG 


Admit Diag 


6 


C 


The principle diagnosis at the time of 
admission. Admitting diagnoses appear on 
hospital claims. 


CADMDT 


Admit Date 


4p 


N 


The date of admission for a hospital claim. 


CAGE 


Age 


2p 


N 


The member's calculated age on the claim date 
of service. Age appears on both claim and 
referral screens. 


CBP 


BP 


4 


C 


A member's assigned benefit package number. 


CCASE 


Case Held 


12 


C 


This field may be used for the following: 

• To store remarks 

• ICD9 codes for ambulatory surgery claims 
can be entered here 

• The baby's diagnosis is entered here for OB J 
delivery claims 



Utilization Support Network 

Harvard Community Health Plan© 



Version 1.1 

September 30. 1993 



Chapter 6: Data Dictionary 



UEUSN Data Dictionary (continued) 



1 UEUSN Report rr 

I Field Name Header Name S,ze Type Description 


CCNTNO 


Contract* 


12 


C 


The member's contract number. Contract 
number is almost always the subscriber's social 
security number. 


CCOUNT 


Count 


2p 


N 


The count of services on a medical claim or the 
number of days on a hospital ciaimiDecimal 
places are not allowed in the count field. 


CDIAG 


Diagnosis 


6 


C 


The ICD9 diagnosis code for medical claims. 
(CDIAG 1 is the equivalent field for hospital 
claims.) 


CDIAG1 


Diagl 


6 


C 


The primary ICD9 discharge diagnosis code for 
hospital claims. 


CDIAG2 


Diag2 


6 


C 


The secondary discharge diagnosis code for 
hospital claims. 


CDIAG3 


Diag3 


6 


C 


The tertiary discharge diagnosis code for 
hospital claims. 


CDIAG4 


Diag4 


6 


C 


Additional discharge diagnosis code for 
hospital claims. 


CDIAG5 


Diag5 


6 


C 


Additional discharge diagnosis code for 
hospital claims. 


CDISDT 


Disch Date 


4p 


N 


The discharge date for hospital claims. (An 
interim claim does not have a discharae date.) 


CDOB 


MbrDOB 


4p 


N 


The member's date of birth. (YYMMDD format) 


CEXCD 


EX Code 


4 


C 


EX codes describe why a claim is pended or 
denied. Code descriptions can be looked up in 
the EX code set in Reference & Controls. 


CFROM 


From Date 


4p 


N 


For medical claims this is the date of service. 
For hospital claims this is the bill start date, (the 
admit date except for interim claims). 


CGROUP 


Group# 


12 


C 


The unique employer group number. 


CHARGD 


Amt Chgd 


4p 


N 


Amount charged appears on claims screens. It 
is the amount charged per service line. 


CHLTCT 


Health Ctr 


12 


C 


The member's HCHP health center code. 
Health center codes are stored in the HC code 
set in Reference & Controls. They are also 
listed on page 7-10. 


CLC 


Location 


4 


C 


The location code identifies the type of 
institution in which a service was performed. 
Location codes are listed on page 7-13. 


CLFLAG 


ClmlstFIg 


1 


N 


This flag is set to one of the following: 

• 1 for the first occurrence of a particular 
claim number 

• for subsequent service lines 



Version 1.1 

September 30. 1993 



Utilization Support Network 

Harvard Community Health Plan© 



Chapter 6: Data Dictionary 



UEUSN Data Dictionary (continued) 



1 UEUSN Report _. . . 
f Field Name " Header Name Size Type Description 


CLMNUM 


Claim# 


12 


C 


Claim numbers consist of the following: 

• The receipt date of the claim in a YYMMDD 
format, (930104). 

• A four digit sequential number. 

The first claim received each day is assigned 
0001. The last claim number indicates the 
number of claims received that day. 
(9301041024 - 1024 claims were received on 
01/04/93.) 


CLST 


ST 


2 


c 


The claim status (paid, denied, pended). Status 
codes are stored in the ST code set in 
Reference & Controls. They are listed on page 
7-22. 


CMDREC 


Med Rec# 


12 


c 


The member's unique HCHP medical record 
number. 


CMEMCL 


CL 


4 


c 


The member's family classification code. These 
codes differentiate members on the same 
contract by identifying the relationship between 
the member and subscriber. The CL code set 
is contained on page 7-7. 


CMEMNO 


Mbr# 


2 


c 


The member number is the two digit extension 
used with the contract number. Member 
numbers identify individual members on a 
contract. (Subscriber = 00, 01 -spouse, 02-first 
child, 03-Second child, etc.) 


CMOD 


Modifier 


2 


c 


Procedure modifiers are used to designate 
assistant surgeons, anesthesia units, or other - 
services that have been altered by some 
special circumstances. These after the original 
procedure code without changing its definition 
or code. Modifier codes are stored in the M1 
code set in Reference & Controls. 


CNAME 


Member Name 


33 


c 


The member's name (first, middle, last). 
"Restricted" appears on employee records. 


CNCOVD 


Covd Days 


2p 


N 


The number of days billed on an inpatient claim. 


COCL 


OCL 


4 


C 


The ILR code that identifies the ieason for an 
ILR pend or adjustment, (e.g., Medicare is 
primary, other insurance is primary, motor 
vehicle accident, etc. . .) ILR codes are used 
for coordination of benefits. They are listed on 
page 7-11. 


CPAIO 


Amt Paid 


4p 


N 


The dollar amount actually paid for a service 
line. I 



Utilization Support Network 

Harvard Community Health Plan© 



Version 1.1 

September 30, 1993 



Chapter 6: Data Dictionary 



UEUSN Data Dictionary (continued) 



1 UEUSN 
> Field Name 


Report 
Header Name 


Size 


Type 


Description 


CPCP 


PCP 


12 


C 


The Costar code, preceded by a P, of the 
member's primary care physician. 


CPDTOT 


Paid Claim Total 


5p 


N 


The total amount paid on a claim. 


CPRDT1 


ICD9 Proc 1 Date 


4p 


N 


The date the primary procedure was 
performed, YYMMDD. (Applies to hospital i 
claims only.) 


CPRDT2 


ICD9 Proc 2 Date 


4p 


N 


The date the secondary procedure was 
performed, YYMMDD. (Applies to hospital 
claims only.) 


CPRDT3 


ICD9 Proc 3 Date 


4p 


N 


The date the tertiary procedure was performed, 
YYMMDD. (Applies to hospital claims only.) 


CPROC 


Procedure 


7 


C 


The primary procedure code. (CPT4, UB82, 
HCPCS procedure codes.) 


CPROC1 


ICD9 Prod 


7 


C 


The primary ICD9.CM procedure code. 
(Applies to hospital claims only.) 


CPROC2 


ICD9 Proc2 


7 


C 


The secondary ICD9.CM procedure code. 
(Applies to hospital claims only.) 


CPROC3 


ICD9 Proc3 


7 


C 


The tertiary ICD9.CM procedure code. (Applies 
to hospital claims only.) 


CSEX 


Sex 


4 


C 


The member's gender. (M - male, F - female) 


CSVUN 


SvcLine* 


2 


C 


The individual service line number within a 
claim. 


CTHRU 


Thru Date 


4p 


N 


For medical claims this is the through service 
date. It is the discharge date on a final bill of a 
hospital claim. For interim bills, it is the thru 
date. 


CTOCHG 


Total Chgs 


5p 


N 


The total charges for all service lines on a claim. 


CTT 


TT 


4 


C 


The treatment type associated with a specific 
procedure code. These codes are stored in the 
TT code set in Reference & Controls. They are 
listed on page 7-24. 


CTYPE 


CP 


2 


C 


The claim type code is one of the following: 

HO - inpatient 

ME - outpatient, SDC, or any professional 

charges 


cuNrrs 


Units 


3 


C 


The number of units for any type of service 
requiring a modifier. 


PAR 


AR 


2 


C 


The referral type code identifies the type of 
service for which a member is being referred, 
(e.g., Inpatient - Obstetrics) AR codes are 
listed on page 7-3. 



Version 1.1 

September 30, 1993 



Utilization Support Network 

Harvard Community Hearth Plan© 



Chapter 6: Data Dictionary 



UEUSN Data Dictionary (continued) 



UEUSN Report _. _ — _ 1 
tr- u m j u Slze T yP e Description 
Field Name Header Name r 


PAS 


AS 


2 


C 


The extent of care code identifies the level of 
approval required for a referral, (e.g., benefit 
coordinator sign-off, clinician etc. . . ) AS codes 
are listed on page 7-5. 


PATTPV 


Att Prov# 


12 


C 


Attending providers only apply to hospital 
referrals. This is the provider number of the 
actual attending provider, if known. 


PATTSP 


Att Prov Sp 


2 


c 


The attending provider's specialty code. 
(Hospital claims only.) 


PCNTNO 


Contract # 


12 


c 


The member's contract number on a referral. 
Contract number is almost always the 
subscriber's contract number. 


PHLTCT 


Health Ctr 


12 


c 


The member's HCHP health center code on a 
referral. Health center codes are stored in the 
HC code set in Reference & Controls. They are 
also listed on page 7-10. 


PMDREC 


MedRec# 


12 


c 


The member's unique HCHP medical record 
number on a referral. 


PMEMNO 


Mbr# 


2 


c 


The member number is the two digit extension 
used with the contract number. Member 
numbers identify individual members on a 
contract. (Subscriber = 00, 01 -spouse, 02-first 
child, 03-Second child, etc.) 


PPRIM 


Prim Diag 


6 


c 


The member's principle diagnosis code when 
the referral was first entered. 


PPROC1 


ICD9 Prod 


7 


c 


Procedure code number 1 from hospital logs. 


PPROC3 


ICD9Proc3 


7 


c 


Procedure code number 3 from hospital logs. 


PREFPV 


Ref Prov# 


12 


c 


On a medical referral this is the HCHP provider 
who ordered/referred a service. On a hospital 
referral it is the admitting provider. 


PREFSP 


Ref Prov SP 


2 


c 


On a medical referral this is the ordering/ 
referring provider's specialty code. On a 
hospital referral it is the admitting provider's 
specialty code. 


PREFTY 


AuttiType 


1 


c 


The referral type code is one of the following: 

• H - Hospital 

• M - Medical 


PRFLAG 


PCISTRg 


1 


N 


This flag is set to one of the following: 

• 1 on first occurrence of a particular referral 

• on subsequent occurrences 


PSITE 


Referral Site 


4 


c 


The three character abbreviation of the site 
from which a referral was generated. 



Utilization Support Network 

Harvard Community Health Plan© 



Version 1 .1 

September 30, 1993 



Chapter 6: Data Dictionary 



UEUSN Data Dictionary (continued) 



1 UEUSN Report 

Field Name Header Name SlZe Type Description 


PSRVPV 


Hosp Prov# 


12 


C 


The facility to which a member has been 
referred. 


PSRVSP 


Hosp SP 


2 


C 


The specialty code of the facility to which a 
member was referred. (For example, HO — -.. 
Acute Care Hospital.) These codes are stored 
in the SP code set in Reference & Controls. 
They are listed beginning on page 7-18. 


PTMPLT 


Template Name 


12 


C 


The template name. Templates authorize the 
services to be performed. (Do not use the H or 
M preceding templates. These are outdated.) 
Templates are listed beginning on page 7-28. 


PTPRC1 


Template Proc 1 


7 


C 


The major or first service authorized by a 
template group of services. Only used on 
medical, ambulance and ER refenals. This 
code is passed to AMRS. 


PTPRC2 


Template Proc 2 


7 


C 


The second service authorized by a template 
group of services. This code is manually input 
by CAG to authorize additional procedures 
because only one template can be used per 
referral. Only used on medical, ambulance and 
ER refenals. 

Example: CTHEAD is used as the template 
and CTABDOMEN/CTPLEVIS are 
also authorized. Cag needs to enter 
these two procedure codes and 
quantities. 


PTPRC3 


Template Proc 3 


7 


C 


The third service authorized by a template 
group of services. (See example above.) 


SRVPS 


ServPS 


4 


C 


On a claim this is the servicing provider status 
code (PS code). It is passed to the referral as 
the referred to status code. PS codes are listed 
on page 7-17. 


SRVPV 


Serv Prov# 


12 


C 


On a claim this is the provider number of the 
vendor providing service to the member. On a 
referral it is the refened to provider code. 


SRVSP 


Serv Prov SP 


2 


C 


On a claim this is the servicing provider's 
specialty code. On a referral it is the refened to 
provider's specialty code. These codes are 
stored in the SP code set in Reference & 
Controls. They are listed on page 7^18. 



Version 1 .1 

September 30, 1 993 



Utilization Support Network 

Harvard Community Health Plan© 



MGD Claims File Layout 



CLINIC HEADER FILE DATA DICTIONARY 



Field 



7ACVST 



12APRV1 



Name 



Number of visits that actually occurred 



Assigned Provider 1 - Copied from the E/B 
file. It is unreliable as a primary provider 
source. Sometimes it is the medical group 
acronym and PCP; example WMAPCP 



19AUTHN 



6AUVST 



25COBTP 



14DIAG2 



Description 



Authorization Number - BLANK 



Number of visits that were authorized-BLANK 



Coordination of Benefits Type 
acronym when applicable 



Parent IPACD 



13DIAG1 Diagnosis KICD-9) - is a required field. 



8ENRSN 



2FORMN 



17GRPID 



29INUSE 



20 IPACD 



31 LOCSV 



15MBAGE 



Encounter reason - HCHP Codes for Type of 
Service - always eg '12' for clinic 



Claim Number- automatically generated by 
the system. It consists of C followed by 7 
numbers. 



Employer Group on the service date 



Record in Use Flag 



Medical Group on the service date 



Location of Service-Medical groups can 
contract with APPs(Affiliated Physician 
Practices). The LOCSV code distinguishes 
whether the service occurred at the medical 
group or at the APP. LOCSV = 3 character 
medical group acronym . 



Diagnosis 20CD-9) - is not a required field. 



Member Age - is calculated based on the 
birth date and the service date to 1/10 of a 
year. 



Type 



Width 



3,0 



3,0 



Format 



6XXX.XX 



6XXX.XX 



3,1 



Start 
Pos 



39 



65 



95 



37 



125 



Range 



71 See ICD-9Code 
book. 



77 See ICD-9 Code 
book. 



41 



See Appendix(ENRSN 
Code Sheet). 



6 



86 



148 



104Range '01' -'L2' See 
Appendix(HCHP Sites 
Code Sheet). 



151 



83 



3MBRNO 



Member Number on the service date - the 
groups enter it and it is verified when the 
claim is processed. 



11 



14 



16MBSEX Member Sex 



280BCHN 



260ENDT 



Original Batch Number- batch number of the 
batch that includes the claim. It is a random 
number entered by the clinical supervisor. 
The first 2 characters represent the COB 
code(parent group of the IPA code). 



85 



'F' OR 'M' 



142 



Claim Entry Date - System Date 



8.0YYYYMMDD 



127 



270ROPR 



Original Entry Operator- operator that loaded 
the tape or diskette to the batch file. 



10 



132 



9PLCSV 



Place of Service Codes - which designate 
where the service took place. In 7/93. HCHP 
switched to Medicare PLCSV codes, which 



are more specific than the HCHP homegrown 



codes were. 



43Range '1V - '99' See 
Appendix(Medicare 
PLCSV Code Sheet). 



CLNHDD.DOC 1 1/29/93 



CLINIC HEADER FILE DATA DICTIONARY 



Field 



Name 



Description 



Type Width 



Format 



Start 
Pos 



Range 



18PLNCD 



Plan Code - Benefit package at the time of 
the claim 



92 



me maim 

Plan Type - First position of the plan code, 
represents the broad category of coverage 

Orrwtirinr* c Af/»Aiint Miimhor - nrn\/iHor'c 



24 PLNTP 



124 



22PRVAC 



Provider's Account Number - provider's 
financial account number or claim number for 
the patient in the provider's data system. 



11 



110 



4PRVNO 



Provider Number - Service Provider, This 
director/ is maintained by the accounting 
department. 



25 See Appendix(MGD 
Sasified Files - 

HPROVP). 

1 Example: 19910101 



1 



PSVDT 



Primary Date of Service - service date 
associated with the claim. 



8,0YYYYMMDD 



30PTCTG 



Physician Category - Linked to the provider 
type, A group category combines more than 
one specialty. 



149 See Appendix(MGD 
Sasified Files - 
HPROVP). 



10PTYPE 



Provider Type - Specialty of the service 
provider example- PHYS-OBGYN for 
obstetricians 



10 



45 See 



Appendix(MGD 
Sasified Files - 
HPROVP). 



5RFPRV 



Referring Provider BLANK, since there was no|C 
referral. 



31 



23SBGRC 



Subgroup Code - NOT USED 



121 



21 



USRFL 



User Field - Always CLNC 



106 



11 



VNDNO 



Vendor Number - Used for accounts payable 
purposes only, NOT APPLICABLE for Clinic 



10 



55 



CLNHDD.DOC 11/29/93 



CLINIC DETAIL FILE DATA DICTIONARY 



Held 



22ADJST 



15AJRSN 



10ALWAM 



27ALWQT 



19APPST 



9BILAM 



23BKTST 



16CMPCD 



11 



COPAM 



2FORMN 



17GLDST 



5MBRNO 



21 



Name 



Adjudication Status - Not Used 



Adjustment Reason 
Schedule 



Allowed Amount - Amount that HCHP allows 
payment for based on the Clinic Fee Schedule 



Allowed Procedure Quantity - not yet used, 
goes with the ALWAM. It will be used by the 
APPs. 



Billed Amount 



Company Code - Not Used 



Copay Amount - Amount of the Member's 
Copay. It is a flat amount paid by the 
enrollee per visit or service regardless of the 
cost of the services provided. - Not Used 



Claim Number- automatically generated by 
the system. It consists of C followed by 7 
numbers. 



General Ledger Distribution Code - Not Used 



NCRSN 



13NETAM 



280BCHN 



240ENDT 



250ROPR 



14PAYST 



4 PCDCD 



Description 



Based on the Clinic Fee 



Accounts Payable Posting Date - Not Used 



Bucket Status - Not Used 



Member Number on the service date - the 
groups enter it and it is verified when the 
claim is processed. 



Not Covered Reason - Not Used 



20NCVAM Not Covered Amount - Not Used 



Net Amount - Portion of the billed amount 
that is reimbursed by HCHP. 



Original Batch Number- batch number of the 
batch that includes the claim. It is a random 
number entered by the clinical supervisor. 
The first 2 characters represent the COB 
code(parert group of the IPA code). 



Claim Entry Date - Date record added 



Original Entry Operator- Original operator that |C 
loaded the tape or diskette to the batch file. 



Pay Status - Always 'X' 



Procedure Code(CPT) 



Type Width Format 



7,2 



7,2 



7,2 



11 



7,2 



7,2 



10 



8,0YYYYMMDD 



Start 
Pos 



95 



73 



56 



118 



84 



52 



96 



75 



60 



76 



28 



93 



89 



68 



112 



97 



102 



72 



19 



Range 



8PCDQT 



Procedure Quantity - The definition of unit of 
service may vary by department. 



3,0 



50 



7POSTD 



Date Record Added 



8,0YYYYMMDD 



14 



6PRVN0 



Provider Number - service provider - This file 
is maintained by the accounting dept. 



39 



1 



PSVDT 



Primary Date of Service - service date 
associated with the claim. 



8,0YYYYMMDD 



1 



Example: 19930101 



18RCVDT 



Receive Date for Claim 
was received. 



Date that the claim 



8,0YYYYMMDD 



79 



3SSVDT 



Specific Date of Service - Same as the 
primary date of service for clinic. 



8.0YYYYMMDD 



14 Example: 19930101 



1 2|WITAM 



Withhold Amount 



7,2 



64 



CLINDDD.DOC 11/29/93 



institutional Header File Data Dictionary 



Field Name 



Description 



Type 



Width 



Format 



POS 



Range 



49ACDYS 



Actual Days - Not Used? 



5,2 



236 



7ACVST 



Length Of Stay 
admit date. 



The discharge date - the 



3.0 



39 



ADMDT 



Admission Date- Date that the patient was 
admitted. 



8,0YYYYMMDD 



1 Example: 19910101 



12APRV1 



Assigned Provider 1 - Copied from the E/B 
file. It is unreliable as a primary provider 
source. Sometimes it is the medical group 
acronym and PCP; example WMAPCP 



65 



24ATPHY 



Attending Physician - Blank - Hospital MD 
codes on the claim form don't match the 
HCHP codes, so they cannot be entered. 



123 



22AUTHN 



Authorization Number - Random number 
generated by the system. It is a unique key. 
One authorization may have only one claim, 
except when a newborn claim is paid off the 
mother's authorization. 



113 



6AUVST 



Number of Visits that were authorized 



3,0 



37 



50 BIRTH W 



Birth Weight - Grams 



4,0 



241 



23CNTRC 



Continuation Record Flag 



122 



42COBTP 



Coordination of Benefits Type 
acronym when applicable 



Parent IPACD 



197 



13DIAG1 



Diagnosis 1QCD-9) 



6XXX.XX 



71 See ICD-9 Code Book. 



14DIAG2 



Diagnosis 2(ICD-9) 



6XXX.XX 



77 See ICD-9 Code Book. 



19DIAG3 



Diagnosis 3QCD-9) 



6XXX.XX 



95 See ICD-9 Code Book. 



30DIAG4 



Diagnosis 4(ICD-9) 



6XXX.XX 



1 54 See ICD-9 Code Book. 



31 DIAG5 



Diagnosis 5QCD-9) 



6XXX.XX 



160 See ICD-9 Code Book. 



26DISDT 



Discharge Date 



8,0YYYYMMDD 



131 



35DISST 



Discharge Status 



1 84 Range '01 ' - '42' See 
Appendix(DISST Code 
Sheet). 



36DRGCD 



DRG Code - Not Used? 



187 



8ENRSN 



Encounter Reason - HCHP codes for type of 
service These codes relate to benefit 
packages, service limitations and exclusions. 



41 See Appendix(ENRSN 
Code Sheet). 



2FORMN 



Claim Number- automatically generated by 
the system. 



8 



17GRPID 



Employer group at the time of the service 



86 



48INUSE 



Record in use flag 



235 



25 IPACD 



Medical group on the service date 



129|Range 'OV-'L^' See 
Appendix(HCHP Sites 
Code Sheet). 



46LCHGD 



47LCHG0 Last change operator 



Last date that the claim was updated 



15MBAGE 



Member Age - is calculated based on the 
birth date and the service date to 1/10 of a 
year. 



3MBRN0 



16MBSEX 



Member Number on the service date - entered C 
into authorization and used when the claim is 
entered. 



Member Sex 



37MEMFL 



Memo Flag 



38MMEDF 



450BCHN 



8,0YYYYMMDD 



10 



3,1 



11 



220 



225 



83 



Major Medical Flag 



Original Batch Number- batch number of the 
batch that includes the claim. 



14 



85 



•F' OR 'M' 



190 



191 



214 



hoshdd.doc 11/30/93 



Institutional Header File Data Dictionary 



Field 



430ENDT 



440R0PR 



Name 



Claim Entry Date 
entered. 



Original Entry Operator - operator that loaded 
the tape to the batch file. 



Description 



Date that the claim is 



Type 



Width 



8,0YYYYMMDD 



10 



Format 



POS 



199 



204 



Range 



9PLCSV 



Place of Service Codes - which designate 
where the service took place. In 7/93. HCHP 
switched to Medicare PLCSV codes, which 



are more specific than the HCHP homegrown 



codes were. 



43 See 



Appendix(Medicare 
PLCSV Code Sheet). 



18PLNCD 



Plan Code - Benefit package at the time of 
the claim 



Plan Type - First position of the plan code, 
represents the broad category of coverage. 

Pntit\/ 



92 



40PLNTP 



193 



41 



PPOEN 



Entity 



39PPROV 



Participating Provider Flag 



194 



192 



28PRVAC 



Provider's Account Number - provider's 
financial account number for the patient in 
the provider's data system. 



11 



4PRVNO 



Provider Number - Service Provider, This 
directory is maintained by the accounting 
department. 



10PTYPE 



Provider Type - Specialty of the service 
provider example- PHYS-OBGYN for 
obstetricians 



10 



5 RFPRV 



Referring Provider- Provider who referred the 
patient for services 



140 



25 See 



Appendix(MGD 
Sasified Files - Hprovl). 



45 



See AppendixIMGD 
Sasified Files - Hprovl). 
Example:the PTYPE for 
hospitals is -INST-HOSPI 



31 



See AppendixfMGD 
Sasified Files -Hprovl). 



29SBGRC 



Subgroup Code 



151 



20SPRC1 



Surgical Procedure Code KICD-9) 



6XX.XX 



21 



SPRC2 



Surgical Procedure Code 20CD-9) 



6XX.XX 



32SPRC3 



Surgical Procedure 30CD-9) 



6XX.XX 



101 See ICD-9 Code Book. 



107 See ICD-9 Code Book. 



166 See ICD-9 Code Book. 



33SPRC4 



Surgical Procedure 40CD-9) 



6XX.XX 



172 See ICD-9 Code Book. 



34SPRC5 



Surgical Procedure 5IICD-9) 



6XX.XX 



178 See ICD-9 Code Book. 



27USRFL 



User Field - INST 



136 



li 



VNDNO 



Vendor Number - Is generally used for 
accounts payable purposes only. However, 
for PRVNO = '999999', (Unknown Provider), it 
can be used to identify the provider. 



10 



55 



hoshdd.doc 11/30/93 



NSTITUTIONAL DETAIL FILE DATA DICTIONARY 












Field 


Name 


Description 


Type 


Width 


Format 


Start 
Pos 


Range 


28 


ADJST 


Adjudication Status -O = not, 1 = manual - Not 
Used 


C 


1 




108 




1 


ADMDT 


Admission Date 


P 


8,0 


YYYYMMDD 


1 


Example: 19910101 


15 


AJRSN 


Adjustment Reason 


C 


2 




73 




10 


ALWAM 


Allowed Amount - Amount that HCHP allows 
payment for based on the Clinic Fee Schedule 


P 


7,2 




56 




25 


APPST 


Accounts Payable Posting Date 


P 


8,0 


YYYYMMDD 


97 




19 


BECAT 


Benefit Category 


c 


3 




81 




9 


BILAM 


Billed Amount - Amount that the vendor billed 
on the claim 


p 


7,2 




52 




29 


BKTST 


Bucket Status, = not counted, 1 = counted - 
Not Used 


c 


1 




109 


to 


16 


CMPCD 


Company Code 


c 


1 




75 




23 


COBPF 


COB Persue Flag 


c 


1 




91 




11 


COPAM 


Copay Amount - Amount of the Member's 
Copay 


p 


7,2 




60 




18 


CPRSN 


Copay Reason 


c 


2 




79 




20 


CRBFL 


Credit Bank Flag 


c 


1 




84 




22 


DCRSN 


Deductible Reason 


c 


2 




89 




21 


DCTAM 


Deduct Amount 


p 


7,2 




85 




2 


FORMN 


Claim Number- automatically generated by 
the system. 


c 


8 




6 




17 


GLDST 


General Ledger Distribution Code 


c 


3 




76 




30 


LCHGO 


Last Change Operator 


c 


10 




110 




5 


MBRNO 


Member Number on the service date - entered 
into authorization and used when the claim is 
entered. 


c 


11 




28 




27 


NCRSN 


Not Covered Reason - Not Used 


c 


2 




106 




26 


NCVAM 


Not Covered Amount - Not Used 


p 


7,2 




102 




13 


NETAM 


Net Amount - Portion of the billed amount 
tthat is reimbursed by HCHP. 


p 


7,2 




68 




14 


PAYST 


Pay Status - Not Posted, Posted or Pended 


c 


1 




720,1, 5,6,7,8,9,N,X,E,F,G 
,H,I,J " 


4 


PCDCD 


Procedure Code(CPT or Revenue or HCPCS) 


c 


9 




19 




8 


PCDQT 


Procedure Quantity 


p 


3,0 




50 




7 


POSTD 


Due Date 


p 


8,0 


YYYYMMDD 


45 




6 


PRVNO 


Provider Number - Service Provider, This 
directory is maintained by the accounting 
department. 


c 


6 




39 


See Appendix(MGD 
Sasified Files - 
Hprovp). 


24 


RCVDT 


Receive Date for Claim 


p 


8,0 


YYYYMMDD 


92 




3 


SSVDT 


Specific Date of Service - Date on which the 
specific procedure occurred. 


p 


8,0 


YYYYMMDD 


14 




12 


WITAM 


Withhold Amount - Medicare Paid Amount 


p 


7,2 




64 





HOSDDD.DOC 1 1/23/93 



REFERRAL HEADER FILE DATA DICTIONARY 



Held 



Name 



Description 



Type 



Width 



Format 



Start 
Pos 



Range 



7ACVST 



Number of Visits that actually occurred. The 
default visit count is 1 . 



3,0 



12APRV1 



Assigned Provider 1 - Copied from the E/B 
file, It is unreliable as a primary provider 
source. Sometimes it is the medical group 
acronym and PCP; example WMAPCP 



39 



65 



19AUTHN 



Authorization Number - Random number 
generated by the system. It is a unique key. 
One authorization may have multiple related 
claims. 



95 



6AUVST 



Number of Visits that were authorized 



3,0 



37 



25 



COBTP 



Coordination of Benefits Type - Parent IPACD 
when applicable 



125 



13DIAG1 



Diagnosis 1QCD-9) 



6xxx.xx 



71 See ICD-9 Code book. 



14DIAG2 



Diagnosis 2(ICD-9) 



6xxx.xx 



77 See ICD-9 Code book. 



8ENRSN 



Encounter Reason - HCHP codes for type of 
service These codes relate to benefit 
packages, service limitations and exclusions. 



41 See Appendix(ENRSN 
Code Sheet). 



2FORMN 



Claim Number 
the system. 



automatically generated by 



17GRPID 



Employer Group on the service date 



8 



86 



29INUSE 



Record in Use Flag 



148 



20 IPACD 



Medical Group on the Service Date 



104 Range '0r-'L2' See 
Appendix(HCHP sites 
Code Sheet). 



15MBAGE 



Member Age - is calculated based on the 
birth date and the service date to 1/10 of a 
year. 



3,1 



83 



3MBRNO 



Member Number on the service date - entered C 
into authorization and used when the claim is 
entered. 



11 



14 



16MBSEX 



Member Sex 



85 



F' OR "M 1 



32NEDCL* 



NED Claim Number- new field added on 
10/17/93, used to key in microfilm number 
for the New England Division PCN claims. 



10 



280BCHN 



Original Batch Number- Batch Number of the 
atch that includes the claim. 



26|OENDT iCIaim Entry Date - Date that the claim was 
ntered. 



8.0YYYYMMDD 



27 OROPR Original Entry Operator 



9 PLCS V 



Place of Service Codes - which designate 
where the service took place. In 7/93. HCHP 



switched to Medicare PLCSV codes, which 



are more specific than the HCHP homegrown 



codes were. 



18PLNCD 



Plan Code - benefit package at the time of 
the claim. 



24PLNTP 



22PRVAC 



Plan Type - First position of the plan code, 
represents the broad category of coverage. 



10 



Provider's Account Number - provider's 
financial account number for the patient in 
the provider's data system. 



156 



142 



1 1 



127 



132 



3 Range '11' - '99'. See 
Appendix(Medicare 
PLCSV Code Sheet). 



92 



124 



110 



REFHDD.DOC 11/29/93 



REFERRAL HEADER FILE DATA DICTIONARY 



Field 



4 PRVNO 



1 



PSVDT 



30PTCTG 



10PTYPE 



31 



21 



Name 



Provider Number - Service Provider, This 
directory is maintained by the accounting 
department. 



Primary Date of Service - Service date 
associated with the claim. 



Physician Category - Linked to the provider 
type, A group category combines more than 
one specialty. 



RECDT 



5RFPRV 



23SBGRC 



USRFL 



1 1 VNDNO 



Description 



Provider Type - Specialty of the service 
provider example- PHYS-OBGYN for 
obstetricians 



Receive Date for Claim - date that the claim 
is received. Claims are stamped with a date 
as they come in. The stamped date is 
entered by the claims operator into the 
RECDT field. 



Referring Provider - Provider who referred the 
patient, is copied over from the Authorization 
file when the claim is entered. 



Subgroup Code - NOT USED 



User Field - RFRL for referral, CLNC for Clinic C 



Vendor Number 
purposes only. 



Used for accounts payable C 



Type Width 



8,0YYYYMMDD 



10 



_4 
10 



Format 



8,0YYYYMMDD 



Start 
Pos 



2 5 See 



AppendixIMGD 
Sasified Files - Hprovp). 



l Example: 19910101 



149 



4 5fSee Appendix(MGD 
Sasified Files - Hprovp). 



151 



121 



106 



55 



Range 



31 See Appendix(MGD 
Sasified Files - Hprovp). 



REFHDD.DOC 11/29/93 



REFERRAL DETAIL DATA DICTIONARY 



Field Name 



Description 



Type 



Width 



Format 



Start 
Pos 



Range 



22ADJST 



Adjudication Status - = not, 1 = manual 



95 



15AJRSN 



Adjustment Reason 



73 



10ALWAM 



Allowed Amount - Amount that HCHP allows 
payment for based on the vendor fee 
schedule) can be manually entered). 



7,2 



56 



19APPST 



Accounts Payable Posting Date 



8,0YYYYMMDD 



84 



9BILAM 



Billed Amount - Amount that the vendor billed 
on the claim 



7,2 



52 



23BKTST 



Bucket Status, O = not counted, 1 = counted - 
Not Used 



96 



16CMPCD 



Company Code 



75 



11 



COPAM 



Copay Amount 
Copay 



Amount of the Member's 



7,2 



60 



25CRLIN 



Clinic/Referral Line- Since all PCN claims are 
entered into the Referral File, this is a way to 
distinguish between PCN claims and MGD 
Claims. C = Clinic and R = Referral. Any claim 
with CRLIN = 'C. is a PCN Clinic claim 



102 



2FORMN 



Claim Number - automatically generated by 
the system. 



8 



17GLDST General Ledger Distribution Code 



76 



5MBRNO 



Member Number on the service date - the 
groups enter it and it is verified when the 
claim is processed. 



11 



28 



21 NCRSN 



Not Covered Reason - Not Used 



93 



20NCVAM Not Covered Amount - Not Used 



7,2 



89 



13NETAM 



Net Amount - Portion of the billed amount 
that is reimbursed by HCHP. 



7,2 



68 



14PAYST 



Pay Status - whether or not the bill has been 
posted or pended. Not posted = 0,posted = 1 . 
If pended, may be C, 6, 7, 9, etc. 



72 



4 PCDCD 



Procedure Code(CPT) 



19 



8PCDQT 



Procedure Quantity - The definition of unit of 
service may vary by department. 



3,0 



50 



7POSTD 



Date Record Added 



8,0YYYYMMDD 



45 



6PRVNO 



Provider Number(PRVNO) Service Provider 



3 9jSee Appendix! MGD 
Sasified Files - Hprovp). 



24 PSLIN 



1 



Primary/Secondary Line - This field is used 
only for PCNs. It indicates whether or not a 
service is primary or secondary. Primary 
services, when performed by a primary care 
physician are covered under the primary care 
capitation, so the claim IS NOT paid. 
Secondary services are not covered under the 
captitation, so the claim is paid. 



PSVDT 



Primary Date of Service - Service date 
associated with the claim. 



8,0YYYYMMDD 



97 p = Primary, 
S = Secondary 
See Appendix(Primarv 
Care Code Sheet). 



lExample: 19910101 



18RCVDT 



_L 



Receive Date for Claim - Date that the claim 
was r eceived. 



8.0YYYYMMDD 



79 



REFDDD.DOC 11/29/93 



REFERRAL DETAIL DATA DICTIONARY 



Held 



3SSVDT 



26UTLAM 



Name 



Specific Date of Service - Date on which the 
specific procedure occurred. 



12WITAM 



Description 



Type Width 



Utilization Amount = ALWAM - Copay, for 
medical groups, this usually equals the net 
amount. For some medical groups, the net 
amount may equal 0, in which case the 
UTLAM should be used. It may not equal the 
net amount for PCNs. 



Withhold Amount - Amount Withheld 



Format 



Start 
Pos 



Range 



8.0YYYYMMDD 



7,2 



7,2 



14 



112 



64 



REFDDD.DOC 11/29/93 




3 BDT5 0DDDbDb5 3